Temporal distribution of motorcyclist injuries and risk of fatalities in relation to age, helmet use, and riding while intoxicated in Khon Kaen, Thailand

Accident Analysis and Prevention 37 (2005) 833–842

Temporal distribution of motorcyclist injuries and risk of fatalities in relation to age, helmet use, and riding while intoxicated in Khon Kaen, Thailand Shinji Nakahara a,∗ , Witaya Chadbunchachai b , Masao Ichikawa a , Nakorn Tipsuntornsak b , Susumu Wakai a a

Department of International Community Health, Graduate School of Medicine, the University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan b Trauma and Critical Care Center, Khon Kaen Regional Hospital, Khon Kaen, Thailand Received 14 January 2005; accepted 2 April 2005

Abstract This study investigated the temporal distribution of risky behaviors among injured motorcyclists, that is, riding unhelmeted or while intoxicated, and showed how they are associated with risk of fatal injuries. Data of motorcyclists injured in Khon Kaen municipality in northeastern Thailand and transferred to Khon Kaen Regional Hospital were obtained from the trauma registry system of the hospital. Case fatalities were compared by time of day, age group, helmet use, and alcohol intoxication. Unhelmeted riding peaked late in the evening and riding while intoxicated peaked around midnight. Both were associated with increased fatality risk after stratification by time of day; the odds ratios were 3.49 (95% confidence interval (CI) = 1.48–9.36) and 3.01 (CI = 1.71–5.19), respectively. Nighttime injuries were not significantly associated with increased fatality risk after stratification by helmet use or alcohol intoxication. Unhelmeted driving was prevalent and associated with higher fatality risk among younger drivers, whereas intoxicated driving was less prevalent among teens but associated with increased risk among those aged 20–39 years. This study shows that riding unhelmeted or while intoxicated can explain the increased fatality risk at night, suggesting that safety education or enforcements should be targeted at specific age groups and appropriate times. © 2005 Elsevier Ltd. All rights reserved. Keywords: Nighttime; Motorcyclist; Helmet; Alcohol

1. Introduction In developing countries, motorcycles are widely used as an economical mode of transportation; however, they also account for the majority of traffic-related morbidity and mortality in these areas. In Thailand, the number of registered motorcycles has been increasing rapidly, from 8.2 million in 1994 to 16.6 million in 2002 (in Khon Kaen: 190,746 in 1994 and 538,264 in 2002) (Alpha Research, 2004), and in 2000, motorcycles were involved in 73% of road traffic crashes and 79% of fatal crashes (Ministry of Public Health, 2002). Although the protective effects of helmets in reducing head injuries and mortality rates have been shown (Gabella ∗

Corresponding author. Tel.: +81 3 5841 3698; fax: +81 3 5841 3422. E-mail address: [email protected] (S. Nakahara).

0001-4575/$ – see front matter © 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.aap.2005.04.001

et al., 1995; Liu et al., 2004; Norvell and Cummings, 2002; Rowland et al., 1996), evaluation studies of mandatory helmet use legislation have also indicated smaller mortality reductions than expected (Graham and Lee, 1986; Ichikawa et al., 2003) or no significant reductions (Sosin and Sacks, 1992). This has led to implication of the risk compensation hypothesis (Adams, 1995; Graham and Lee, 1986); however, an alternative explanation is possible. Evaluation studies that focus on post-law behavior changes during the daytime, such as increased helmet use, differ from those during the nighttime when the majority of motorcycle crashes occur (Doyle et al., 1995; Panichaphongse et al., 1995; Sirathranont and Kasantikul, 2003). There is a higher risk of severe injuries with nighttime motorcycle crashes than those that occur in the daytime (Cirera et al., 2001; Lin et al., 2003a; Quddus et al., 2002; Valent et al.,

834

S. Nakahara et al. / Accident Analysis and Prevention 37 (2005) 833–842

2002). This might be because high-risk behaviors are more prevalent at night; riding unhelmeted (Skalkidou et al., 1999) and while intoxicated (Hingson and Winter, 2003; Kasantikul et al., 2005) are more common at night, and both are especially prevalent in developing countries where law enforcement activities are weaker at night (Chongsuvivatwong et al., 1999; Conrad et al., 1996; Ichikawa et al., 2003). These behaviors also tend to be linked; among injured intoxicated motorcyclists, unhelmeted riding, speeding, and crashes at night were common (Kasantikul et al., 2005; Luna et al., 1984; Nelson et al., 1992; Peek-Asa and Kraus, 1996; Shankar and Mannering, 1996). Intoxicated driving (Kasantikul et al., 2005; Lin et al., 2003a; Luna et al., 1984; Shankar and Mannering, 1996) and speeding (Lin et al., 2003a; Peek-Asa and Kraus, 1996) are also known to increase the severity of injuries associated with motorcycle accidents. Further, younger people in their teens or early twenties, who represent the majority of injured motorcyclists (Bray et al., 1985; Doyle et al., 1995; Gabella et al., 1995; Norvell and Cummings, 2002; Swaddiwudhipong et al., 1994), tend to partake in high-risk behaviors when driving at night (Williams, 2003). It has been shown that younger car drivers are also at higher risk of fatal or non-fatal crashes at nighttime (Williams, 2003; Rice et al., 2003), resulting in countermeasures such as provisional licensing schemes (Lin and Fearn, 2003). However, information is lacking with regard to how these factors are associated with increased risk of fatal injuries among motorcyclists at night. Using data from the trauma registry of Khon Kaen Regional Hospital in Thailand, this study investigated temporal changes in riding unhelmeted and while intoxicated and their influence on risk of fatal injuries among injured motorcyclists, paying particular attention to different age groups.

2. Methods 2.1. Study setting This study investigated motorcyclists injured in Khon Kaen municipality and transferred or referred to the Trauma Center of Khon Kaen Regional Hospital between 1998 and 2002. Data were derived from the trauma registry of the center. This center provides tertiary trauma care covering the Khon Kaen Province; details of the center and trauma registry are described elsewhere (Ichikawa et al., 2003). Khon Kaen Province, which has a population of more than 1.7 million, is located in northeastern Thailand where enforcement efforts based on the Helmet Act started on January 1, 1996. Consequently, helmet use in the daytime has reached almost 100%, though helmet use at night has showed only a slight increase. Emergency room staff at Khon Kaen Regional Hospital collect information on age, the time of a crash, alcohol consumption, and helmet use by interviewing patients at the time of admission. If the patient is unable to answer, information

is collected from family members or those who brought the individual to the hospital. Blood alcohol levels are not measured, but alcohol consumption is determined based on the interview and smell immediately after arrival at the emergency room. Although this measurement seems subjective, a study in Thailand described the tendency of Thai motorists to freely disclose their unfavorable driving behaviors, such as driving while intoxicated, after involvement in crashes (Kasantikul et al., 2005). Emergency room nurses are in charge of collecting clinical information data. Fatal cases include deaths that occur before arrival or while in the emergency room, and all deaths before discharge. Since those who die at the scene of the crash are brought to the Forensic Department of Khon Kaen University, we were not able to access data of such cases in the present study. 2.2. Analysis We examined the temporal distribution of injuries among motorcyclists and case fatalities according to helmet use and alcohol consumption. To reduce the influence of random variations, figures show 5 h moving averages of injured cases; however, for the temporal distribution of case fatalities, 7 h averages were used because the number of fatal cases was so small that fluctuations were greater. We then examined associations between the risk of death among injured motorcyclists and helmet use, alcohol consumption, and time of day. Odds ratios (ORs) were used to estimate the risk of death while riding unhelmeted, intoxicated or at night, compared to that while riding helmeted, sober or in the daytime. We were only able to examine the number of severely injured motorcyclists treated in the trauma center; those with less severe injuries who received care at lower levels or who were not injured were not included in the present study. Nighttime was defined as from 18:01 to 06:00. As an outcome measure, we calculated case fatalities, rather than mortality and morbidity rates, using population, the number of registered motorcycles or vehicle km traveled as the denominator, because although most severe cases are transferred to the Trauma Center, Khon Kaen Regional Hospital is not the only tertiary care hospital in the study area. Since risk-taking behaviors, such as riding unhelmeted, while intoxicated or at nighttime, are likely to be linked and thus to confound each other, and since it is likely that they differ by age, we conducted stratified analysis by time of day, helmet use, intoxication, and age. We also calculated overall odds ratios (Mantel-Haenszel odds ratios; MH ORs) adjusted for the confounding factors. If the MH ORs were nearer to the null value (one) than the crude odds ratios, confounding was considered likely. Since age was considered an effect modifier, we did not calculate the MH ORs for stratification by age. We used a statistical software package (Epi-Info; Center for Disease Control and Prevention, Atlanta, Ga) to calculate the odds ratios and 95% confidence intervals. When a cell with a value less than 5 was obtained, exact confidence limits for crude odds ratios were calculated.

S. Nakahara et al. / Accident Analysis and Prevention 37 (2005) 833–842

835

18:01 to 06:00 were more likely than fatal daytime injuries (OR = 1.73).

3. Results 3.1. Motorcyclist characteristics

3.2. Temporal distribution of motorcyclist injuries according to helmet use and intoxication

Of the 9948 injured motorcyclist cases examined, helmet use and alcohol consumption were unknown among 191 and 205, respectively (Table 1); these cases were excluded from analyses where helmet use or alcohol consumption was considered. Male motorcyclists accounted for 74.7% of all cases and were more likely to experience fatal injuries than females (OR = 2.04). Younger motorcyclists aged less than 20 years and those in their twenties accounted for 31.8 and 41.1%, respectively, while those aged 40 years or older accounted for 11.2%. Younger motorcyclists aged less than 20 years tended to show lower case fatalities than other age groups, though the difference was not statistically significant (p = 0.073). Case fatalities across years (p = 0.436) did not differ significantly. Unhelmeted motorcyclists accounted for 74.9% and had a higher risk of fatal injuries than helmeted drivers (OR = 3.87). Intoxicated motorcyclists represented 36.5% and had a higher risk of fatal injuries than sober individuals (OR = 2.97). Occurrence of motorcyclist injuries was highest from 18:01 to 24:00 (33.2%), and fatal injuries during the nighttime from

The number of unhelmeted motorcyclists increased in the afternoon and peaked in the evening; numbers of intoxicated motorcyclists also increased in the afternoon but peaked at midnight. Unhelmeted sober individuals peaked in the evening and decreased before midnight, while unhelmeted intoxicated individuals were rare in the daytime but highly prevalent during the nighttime (Fig. 1). Helmeted sober motorcyclists increased during the daytime with two peaks in the morning and evening, respectively, and helmeted intoxicated individuals were rare throughout the day but showed a slight increase at night. 3.3. Fatal risk according to time, helmet use, and alcohol consumption: stratified analyses When stratified by time of day, unhelmeted and intoxicated motorcyclists showed a more than two-fold higher risk

Table 1 Characteristics of motorcyclists injured in Khon Kaen municipality and case fatalities from 1998 to 2002 Distribution

Fatal cases (n)

Case-fatality (%)

ORa (95% CIb ) or χ2 and p-value

n

%

Sex M F

7435 2513

74.7 25.3

90 15

1.21 0.60

2.04 (1.15, 3.68)

Age 10–19 years 20–29 years 30–39 years 40 years or over

3162 4093 1578 1115

31.8 41.1 15.9 11.2

22 46 23 14

0.70 1.12 1.46 1.26

χ2 = 6.97; p = 0.073

Year 1998 1999 2000 2001 2002

1865 2074 2000 1934 2075

18.7 20.8 20.1 19.4 20.9

23 20 27 18 17

1.23 0.96 1.35 0.93 0.82

χ2 = 3.79; p = 0.436

Helmet use Helmet (+) Helmet (−) Unknown

2445 7312 191

25.1 74.9

6 69 30

0.25 0.94 15.71

3.87 (1.62, 9.90)

Alcohol intoxication Alcohol (+) Alcohol (−) Unknown

3554 6189 205

36.5 63.5

44 26 35

1.24 0.42 17.07

2.97 (1.78, 4.97)

Time of crash 0:01–6:00 6:01–12:00 12:01–18:00 18:01–24:00

2047 2097 2504 3300

20.6 21.1 25.2 33.2

29 12 23 41

1.42 0.57 0.92 1.24

χ2 = 8.80; p = 0.032

4601 5347

46.3 53.7

35 70

0.76 1.31

1.73 (1.13, 2.66)

Daytime (6:01–18:00) Nighttime (18:01–6:00) a b

Odds ratio. Confidence interval.

836

S. Nakahara et al. / Accident Analysis and Prevention 37 (2005) 833–842

Fig. 1. Temporal distribution of motorcyclist injuries (fatal and non-fatal) in Khon Kaen municipality according to helmet use and alcohol consumption between 1998 and 2002 (five-hour moving average).

of fatal crashes than helmeted and sober individuals in each stratum (Table 2). After adjusting for time of day, as shown by the MH OR, unhelmeted and intoxicated motorcyclists showed significantly higher risks of fatal crashes; 3.49 and 3.01 times, respectively. On the other hand, when stratified by helmet use or alcohol consumption, the risk of fatal injuries did not differ between daytime and nighttime; the MH ORs were 1.25 and 0.99, respectively. Intoxicated unhelmeted motorcyclists showed the highest risk of fatal injuries throughout the day (Fig. 2). Unhelmeted sober motorcyclists showed the second highest risk with two peaks in the early morning and evening, respectively. Helmeted drivers, whether sober or intoxicated, showed a lower risk than unhelmeted drivers throughout the daytime until midnight. Helmeted intoxicated drivers showed a high risk in the early morning. To separate the effects of unhelmeted and intoxicated riding, we performed stratified analyses by helmet use and intoxication (Table 3). Among unhelmeted motorcyclists, riding while intoxicated had a two-fold higher risk of fatal injuries than riding while sober (OR = 2.24). Among helmeted motorcyclists, the risk of riding while intoxicated did not differ from that of riding while sober. However, after adjusting for helmet use, riding while intoxicated showed a significantly higher risk (MH OR = 2.16). When stratified by intoxication, riding while unhelmeted resulted in a more than two-fold higher risk

than riding while helmeted among both intoxicated and sober individuals (OR = 5.24 and 2.63, respectively); however, this was not statistically significant. After adjusting for intoxication, riding unhelmeted showed a significantly higher fatality risk (MH OR = 3.26). 3.4. Unhelmeted and intoxicated motorcyclists according to age Among motorcyclists aged less than 20 years and between 20 and 29 years, occurrence of injuries peaked at 20:01–21:00 and remained high until early morning (Fig. 3). Among those aged 30–39 years and 40 or over, two low peaks in the morning and evening, respectively, were seen; occurrence of injuries at midnight and in the early morning was low in these age groups. Riding unhelmeted was common among younger age groups, accounting for 86.9 and 75.6% of injuries among those aged less than 20 years and between 20 and 29, respectively. In older groups, the proportion of those riding unhelmeted was lower in the daytime. In contrast, riding while intoxicated was not common even in the nighttime among those aged less than 20, whereas it was common among those aged 20–29 and 30–39. Among motorcyclists aged less than 30 years, riding unhelmeted and while intoxicated peaked in the evening and at midnight, respectively, while both peaked in the evening among those aged 30 years or over. Intoxicated

S. Nakahara et al. / Accident Analysis and Prevention 37 (2005) 833–842

837

Table 2 Effects of helmet use and alcohol consumption on fatal risks among motorcyclists injured in Khon Kaen municipality by time of day from 1998 to 2002 Total (n)

Fatal cases (n)

Case-fatality (%)

ORa (95%CIb )

Daytime Helmet (+) Helmet (−)

1704 2832

3 24

0.18 0.85

4.85 (1.39, 20.21)d

Nighttime Helmet (+) Helmet (−)

741 4480

3 45

0.40 1.00

2.50 (0.74, 10.08)d

Daytime Alcohol (+) Alcohol (−)

838 3697

10 16

1.19 0.43

2.78 (1.17, 6.50)

Nighttime Alcohol (+) Alcohol (−)

2716 2492

34 10

1.25 0.40

3.15 (1.49, 6.81)

Alcohol (+) Daytime Nighttime

838 2716

10 34

1.19 1.25

1.05 (0.50, 2.28)

Alcohol (−) Daytime Nighttime

3697 2492

16 10

0.43 0.40

0.93 (0.39, 2.16)

Helmet (+) Daytime Nighttime

1704 741

3 3

0.18 0.40

2.30 (0.31, 17.24)e

Helmet (−) Daytime Nighttime

2832 4480

24 45

0.85 1.00

1.19 (0.70, 2.01)

a b c d e

M-H ORc (95% CIb )

3.49 (1.48, 9.36)

3.01 (1.71, 5.19)

0.99 (0.57, 1.73)

1.25 (0.76, 2.09)

Odd ratio. Confidence interval. Mantel-Haenszel odds ratio. Due to large number of cases, exact limits could not be calculated. Exact limits were calculated.

motorcyclists accounted for 23.0% of those aged less than 20 years, 45.4% of those aged 20–29 years, 42.4% of those aged 30–39 years, and 33.9% of those aged 40 or over.

4. Discussion

3.5. Risk of fatal injuries according to age and time

The present study showed that riding unhelmeted and while intoxicated were both associated with higher risk of fatal injuries after stratification by time of day, whereas the risk at night did not differ from that during the daytime after stratification by riding unhelmeted or while intoxicated. This suggests that the increased risk of fatal injuries at night is mainly attributed to riding unhelmeted and while intoxicated, independent of nighttime road conditions or other high-risk riding behaviors. In addition, both riding unhelmeted and while intoxicated increased the fatal risk independent of the effect of each other. Both riding unhelmeted and while intoxicated increased during the nighttime, but they showed different patterns. Crashes among intoxicated motorcyclists peaked around midnight, consistent with a previous study in Thailand (Kasantikul et al., 2005), whereas those among unhelmeted motorcyclists peaked in the evening (Fig. 1). This reflects the crash patterns among young motorcyclists aged less than 30 years who account for the majority of injured cases. These crash patterns are different from those of motorcy-

Younger motorcyclists accounted for the majority of nighttime crashes: teens and those aged 20–29 years represented 33.1 and 44.8%, respectively. Among teens, nighttime crashes were not associated with an increased risk of fatal injuries (OR = 0.94); however, they were associated with an approximate two-fold higher risk of fatal injuries among those aged 20–29 years (OR = 2.27), 30–39 years (OR = 2.11), and 40 years or older (OR = 2.11), though only results of those aged 20–29 years showed statistical significance (Table 4). Alcohol use was not associated with fatal injuries among those aged less than 20 years; however, in contrast, while none of the helmeted motorcyclists experienced fatalities, 0.6% of unhelmeted motorcyclists did so. Both riding unhelmeted and while intoxicated resulted in a significantly increased risk of death among those aged 20–39 years: ORs ranged from 2.97 to 6.92. Riding unhelmeted or while intoxicated tended to result in a higher risk of death among those aged 40 years or over, though this was not statistically significant.

4.1. High-risk behaviors at night

838

S. Nakahara et al. / Accident Analysis and Prevention 37 (2005) 833–842

Fig. 2. Case fatalities among motorcyclists injured in Khon Kaen municipality according to helmet use and alcohol consumption (seven-hour moving average).

clists aged 30 years or over (Fig. 3). Previous studies have shown that high risk driving behaviors among teens take place after 10 p.m. (Williams, 2003); however, such details among motorcyclists have not been well described. The presented information concerning different peaks of unhelmeted and intoxicated motorcyclists will therefore be helpful in devising specific countermeasures against such risky behaviors.

The relatively high prevalence of riding unhelmeted and while intoxicated at night among injured motorcyclists is perhaps attributable to insufficient law enforcement at night. In a questionnaire survey in Indonesia, respondents reported that they were less likely to wear helmets at night when there were no police officers on the road (Conrad et al., 1996). In Thailand, after implementation of the motorcycle helmet law in 1996, helmet use during the daytime reached almost

Table 3 Effects of helmet use adjusted for alcohol consumption and effects of alcohol consumption adjusted for helmet use among injured motorcyclists in Khon Kaen municipality from 1998 to 2002 Total cases (n)

Fatal cases (n)

Case-fatality (%)

ORa (95% CIb )

Helmet (−) Alcohol (+) Alcohol (−)

3064 4165

36 22

1.2 0.5

2.24 (1.28, 3.94)

Helmet (+) Alcohol (+) Alcohol (−)

442 1985

1 4

0.2 0.2

1.12 (0.02, 11.38)d

Alcohol (+) Helmet (+) Helmet (−)

442 3064

1 36

0.2 1.2

5.24 (0.88, 213.3)d

Alcohol (−) Helmet (+) Helmet (−)

1985 4165

4 22

0.2 0.5

2.63 (0.86, 9.01)e

a b c d e

Odds ratio. Confidence interval. Mantel-Haenszel odds ratio. Exact limits were calculated. Due to large number of cases, exact limits could not be calculated.

M-H ORc (95% CIb )

2.16 (1.25, 3.68)

3.26 (1.21, 9.16)

S. Nakahara et al. / Accident Analysis and Prevention 37 (2005) 833–842

839

Fig. 3. Temporal distribution of unhelmeted and intoxicated riding among motorcyclists injured (non-fatal and fatal cases) in Khon Kaen municipality according to age group between 1998 and 2002 (five-hour moving average).

840

S. Nakahara et al. / Accident Analysis and Prevention 37 (2005) 833–842

Table 4 Effects of time of day, helmet use, and alcohol consumption on fatal risks among motorcyclists by age in Khon Kaen municipality from 1998 to 2000 Total cases (n)

Fatal cases (n)

Case-fatality (%)

ORa (95% CIb )

10–19 years Daytime Nighttime

1392 1770

10 12

0.72 0.68

0.94 (0.38, 2.36)

20–29 years Daytime Nighttime

1696 2397

11 35

0.65 1.46

2.27 (1.11,4.76)

30–39 years Daytime Nighttime

832 746

8 15

0.96 2.01

2.11 (0.84, 5.47)

40 years or over Daytime Nighttime

681 434

6 8

0.88 1.84

2.11 (0.66, 6.90)

10–19 years Helmet (+) Helmet (−)

407 2709

0 16

0.00 0.59

–

20–29 years Helmet (+) Helmet (−)

979 3036

2 31

0.20 1.02

5.04 (1.18, 30.49)d

30–39 years Helmet (+) Helmet (−)

578 961

2 15

0.35 1.56

4.57 (1.06, 41.27)c

40 years or over Helmet (+) Helmet (−)

481 606

2 7

0.42 1.16

2.80 (0.53, 27.71)c

10–19 years Alcohol (+) Alcohol (−)

713 2390

3 10

0.42 0.42

1.01 (0.22, 3.95)c

20–29 years Alcohol (+) Alcohol (−)

1817 2189

22 9

1.21 0.41

2.97 (1.30,6.96)

30–39 years Alcohol (+) Alcohol (−)

655 889

15 3

2.29 0.34

6.92 (1.88, 30.17)c

40 years or over Alcohol (+) Alcohol (−)

369 721

4 4

1.08 0.55

1.96 (0.41, 9.37)c

a b c d

Odds ratio. Confidence interval. Exact limits were calculated. Due to large number of cases, exact limits could not be calculated.

100% whereas use at night remained as low as 16–28% (Chadbunchachai, 2002). In addition, alcohol consumption is more common at night when drinking establishments are open. Public transportation systems are not well established in Thailand, and as a result, intoxicated motorcyclists are most often involved in crashes on their way home from bars or restaurants (Kasantikul et al., 2005). 4.2. Difference according to age No increased risk of fatal crashes at night was found among teens despite the fact that riding unhelmeted was prevalent at nighttime. One reason might be that riding unhelmeted among teens was as prevalent in the daytime as at nighttime, resulting in only a slight difference in fatality

risk between day and night. Another reason is that in this age group compared to other age groups, riding while intoxicated was relatively uncommon even in the nighttime. This finding is consistent with a study in Australia where, of injured motorcyclists admitted to hospital, those aged 20–49 were more than twice as likely as teens to have a blood alcohol concentration higher than 80 mg/dl (Holubowycz and McLean, 1995). This reduced intoxicated riding among teens, despite their tendency for risky behaviors, might have resulted from less frequent drinking habits. A survey in Thailand showed that drinking frequency was lower among younger binge drinkers, whereas it did not differ by age among infrequent drinkers (Assanangkornchai et al., 2000). According to the Ministry of Public Health (2002), Thailand, the proportion of male alco-

S. Nakahara et al. / Accident Analysis and Prevention 37 (2005) 833–842

hol drinkers peaks at an age of 35–39 years. Further, alcohol problems were less frequent among teenage emergency room patients compared to those aged 20–49 in regional hospitals in Thailand (Lapham et al., 1998). 4.3. Limitations We could not control for various determinants of injury severity such as speed, darkness, and traffic situations at the time of the crash, all of which vary by time of day (Lin et al., 2003a; Shankar and Mannering, 1996; Valent et al., 2002), because the hospital database did not include detailed information on these parameters. However, these factors were taken into account by stratification by time of day, even though we could not assess their individual effects: the difference in fatal risk between daytime and nighttime crashes after stratification by helmet and alcohol use indicates effects of these factors as a whole independent of helmet or alcohol use. However, the stratified analyses resulted in no differences between the daytime and nighttime risk of fatal crashes. A possible explanation for this is that high-risk behaviors are related; intoxicated motorcyclists tend to drive faster while not wearing helmets (Luna et al., 1984; Nelson et al., 1992; Peek-Asa and Kraus, 1996; Shankar and Mannering, 1996). There is also the possibility of selection bias, which might lead to lower case fatalities at night. Non-fatally injured motorcyclists showing less severe injuries, who would normally have gone to lower level hospitals or clinics in the daytime, might be admitted to the regional hospital during the nighttime when lower level hospitals are closed. Further, those who died at the scene of the crash were not included in the present study. From our findings, we can assume that such motorcyclists were also likely to be unhelmeted and intoxicated and to be involved in crashes at night. These biases would increase the denominator and decrease the numerator of the case fatality calculations. Although this would result in underestimations of the difference between daytime and nighttime, could the results still indicate the significantly higher fatal risk at night mainly due to riding unhelmeted and while intoxicated. Those whose alcohol consumption or helmet use was unknown would have over-represented fatal cases, probably because of difficulties in collecting data from severely injured patients. This could have caused overestimations of risk if such individuals were more likely than other motorcyclists to be helmeted and sober, which is unlikely. Instead, they are likely to take riskier behaviors resulting in fatal crashes. In addition, they accounted for only a small proportion, suggesting minimal selection bias, if at all. The stratification yielded too few fatal cases in some categories resulting in wide confidence intervals. Although this might have resulted in relatively inaccurate risk estimates, our main findings had statistical significance. Increasing the sample size would narrow the confidence intervals leading to more accurate estimations, but would not alter the statistical significance of the obtained results.

841

Although the information on alcohol consumption and helmet use was based on interviews, underestimates of alcohol consumption or overestimates of helmet use are unlikely given the high prevalence of motorcyclists who rode unhelmeted or while intoxicated among the study population compared to the general population (Chadbunchachai, 2002; Chongsuvivatwong et al., 1999). The method used to determine alcohol consumption seems subjective and less accurate compared with blood or breath tests; however, limited resources do not allow testing of all trauma patients and it is believed that the above method did not distort the results. Kasantikul et al. (2005) indicated in their study examining the effect of alcohol on motorcycle crashes, where one third of riders were tested for blood alcohol concentrations and others were interviewed, that both non-tested and tested drinkers were significantly different from non-drinkers in crash characteristics. We admit the possibility of misdiagnosis of drinkers as non-drinkers and vice versa, which would produce biases toward null values. Despite such biases, however, riding unhelmeted or while intoxicated had a significantly higher risk of fatal injuries. The high proportion of motorcyclists riding unhelmeted or while intoxicated among the injured motorcyclists included in this study cannot be used to represent general motorcyclists driving in Khon Kaen municipality. Motorcyclists who take risks, such as riding unhelmeted or while intoxicated, are at higher risk of crashes (Lin et al., 2003b; Pitaktong et al., 2004), and the prevalence of such riding behaviors should be lower among general motorcyclists. However, we can assume that differences in the prevalence of such behaviors by age and time of day among the injured reflect behavioral differences in the general population. That is, the increased prevalence of unhelmeted and intoxicated riding at night among the injured motorcyclists does reflect the increased prevalence of such behaviors at night among motorcyclists in this region in general; previous studies have indicated that such behaviors increase late at night (Chadbunchachai, 2002; Chongsuvivatwong et al., 1999).

5. Conclusions Although further research using more representative subjects, including those who died at the scene and those with less severe injuries, is required, the findings suggest that higher nighttime prevalence of riding unhelmeted and while intoxicated explain the increased fatal risk among injured motorcyclists at night in Khon Kaen. Countermeasures should therefore include augmentation of law enforcement at night, targeted at those riding unhelmeted in the evening and while intoxicated around midnight. We also indicated different patterns of crash-related behaviors according to age, which could be useful in designing intervention measures targeted at each age group. The daytime-nighttime gap is a plausible explanation for why helmet use legislations might not achieve expected

842

S. Nakahara et al. / Accident Analysis and Prevention 37 (2005) 833–842

mortality reductions. Observations of daytime behavior changes, that is, increased helmet use, seem to show the impact of such legislations; however, examinations of the mortality change that reflects the reduced use of helmets at night, when the majority of motorcycle crashes occur, suggest that the reduction is smaller than the prediction. To evaluate the impact more accurately, we therefore need to examine mortality changes with regard to time of day.

Acknowledgements We would like to thank the director of Khon Kaen Regional Hospital, Dr. Vithya Jarupoonphol, for his cooperation and support. This study was partially supported by a Grant-inAid for Young Scientists B (15790296) from the Ministry of Education, Culture, Sports, Science and Technology, Japan.

References Adams, J.G.U., 1995. Risk. UCL Press, London. Alpha Research Co Ltd., 2004. Thailand in Figures, 2003–2004. Alpha Research Co., Ltd, Bangkok. Assanangkornchai, S., Saunders, J.B., Conigrave, K.M., 2000. Patterns of drinking in Thai men. Alcohol Alcohol 35 (3), 263–269. Bray, T., Szabo, R., Timmerman, L., Yen, L., Madison, M., 1985. Cost of orthopedic injuries sustained in motorcycle accidents. JAMA 254 (17), 2452–2453. Chadbunchachai, W. (Ed.), 2002. Development Lessons: Injury Prevention and Control Khon Kaen Case Study. Khon Kaen Regional Hospital, Khon Kaen. Chongsuvivatwong, V., Ritsmitchai, S., Suriyawongpaisal, P., Chariyalertsak, S., Kosuwan, W., Punyaratabandhu, P., Sutiwipakorn, W., 1999. High prevalence of drink-driving in Thailand. Drug Alcohol Rev. 18 (3), 293–298. Cirera, E., Plasencia, A., Ferrando, J., Segui-Gomez, M., 2001. Factors associated with severity and hospital admission of motor-vehicle injury cases in a southern European urban area. Eur. J. Epidemiol. 17 (3), 201–208. Conrad, P., Bradshaw, Y.S., Lamsudin, R., Kasniyah, N., Costello, C., 1996. Helmets, injuries and cultural definitions: motorcycle injury in urban Indonesia. Accid. Anal. Prev. 28 (2), 193–200. Doyle, D., Muir, M., Chinn, B., 1995. Motorcycle accidents in Strathclyde Region, Scotland during 1992: a study of the injuries sustained. Health Bull. (Edinb). 53 (6), 386–394. Gabella, B., Reiner, K.L., Hoffman, R.E., Cook, M., Stallones, L., 1995. Relationship of helmet use and head injuries among motorcycle crash victims in El Paso County, Colorado, 1989–1990. Accid. Anal. Prev. 27 (3), 363–369. Graham, J.D., Lee, Y.H., 1986. Behavioral-response to safety regulation—the case of motorcycle helmet-wearing legislation. Policy Sci. 19 (3), 253–273. Hingson, R., Winter, M., 2003. Epidemiology and consequences of drinking and driving. Alcohol Res. Health 27 (1), 63–78. Holubowycz, O.T., McLean, A.J., 1995. Demographic characteristics, drinking patterns and drink-driving behavior of injured male drivers and motorcycle riders. J. Stud. Alcohol. 56 (5), 513–521. Ichikawa, M., Chadbunchachai, W., Marui, E., 2003. Effect of the helmet act for motorcyclists in Thailand. Accid. Anal. Prev. 35 (2), 183–189. Kasantikul, V., Ouellet, J.V., Smith, T., Sirathranont, J., Panichabhongse, V., 2005. The role of alcohol in Thailand motorcycle crashes. Accid. Anal. Prev. 37 (2), 357–366.

Lapham, S.C., Skipper, B.J., Brown, P., Chadbunchachai, W., Suriyawongpaisal, P., Paisarnsilp, S., 1998. Prevalence of alcohol problems among emergency room patients in Thailand. Addiction 93 (8), 1231–1239. Lin, M.L., Fearn, K.T., 2003. The provisional license: nighttime and passenger restrictions—a literature review. J. Safety Res. 34 (1), 51–61. Lin, M.R., Chang, S.H., Huang, W., Hwang, H.F., Pai, L., 2003a. Factors associated with severity of motorcycle injuries among young adult riders. Ann. Emerg. Med. 41 (6), 783–791. Lin, M.R., Chang, S.H., Pai, L., Keyl, P.M., 2003b. A longitudinal study of risk factors for motorcycle crashes among junior college students in Taiwan. Accid. Anal. Prev. 35 (2), 243–252. Liu, B., Ivers, R., Norton, R., Blows, S., Lo, S.K., 2004. Helmets for preventing injury in motorcycle riders. Cochrane Database Syst. Rev. 2, CD004333. Luna, G.K., Maier, R.V., Sowder, L., Copass, M.K., Oreskovich, M.R., 1984. The influence of ethanol intoxication on outcome of injured motorcyclists. J. Trauma 24 (8), 695–700. Ministry of Public Health, 2002. Thailand health profile, 1999–2000. Ministry of Public Health, Bangkok, pp.220–229. Nelson, D., Sklar, D., Skipper, B., McFeeley, P.J., 1992. Motorcycle fatalities in New Mexico: the association of helmet nonuse with alcohol intoxication. Ann. Emerg. Med. 21 (3), 279–283. Norvell, D.C., Cummings, P., 2002. Association of helmet use with death in motorcycle crashes: a matched-pair cohort study. Am. J. Epidemiol. 156 (5), 483–487. Panichaphongse, V., Watanakajorn, T., Kasantikul, V., 1995. Effects of law promulgation for compulsory use of protective helmets on death following motorcycle accidents. J. Med. Assoc. Thai. 78 (10), 521– 525. Peek-Asa, C., Kraus, J.F., 1996. Alcohol use, driver, and crash characteristics among injured motorcycle drivers. J. Trauma 41 (6), 989–993. Pitaktong, U., Manopaiboon, C., Kilmarx, P.H., Jeeyapant, S., Jenkins, R., Tappero, J., Uthaivoravit, W., van Griensven, F., 2004. Motorcycle helmet use and related risk behaviors among adolescents and young adults in Northern Thailand. Southeast Asian J. Trop. Med. Public Health 35 (1), 232–241. Quddus, M.A., Noland, R.B., Chin, H.C., 2002. An analysis of motorcycle injury and vehicle damage severity using ordered probit models. J. Safety Res. 33 (4), 445–462. Rice, T.M., Peek-Asa, C., Kraus, J.F., 2003. Nighttime driving, passenger transport, and injury crash rates of young drivers. Inj. Prev. 9 (3), 245–250. Rowland, J., Rivara, F., Salzberg, P., Soderberg, R., Maier, R., Koepsell, T., 1996. Motorcycle helmet use and injury outcome and hospitalization costs from crashes in Washington State. Am. J. Public Health 86 (1), 41–45. Shankar, V., Mannering, F., 1996. An exploratory multinomial logit analysis of single-vehicle motorcycle accident severity. J. Safety Res. 27 (3), 183–194. Sirathranont, J., Kasantikul, V., 2003. Mortality and injury from motorcycle collisions in Phetchaburi Province. J. Med. Assoc. Thai. 86 (2), 97–102. Skalkidou, A., Petridou, E., Papadopoulos, F.C., Dessypris, N., Trichopoulos, D., 1999. Factors affecting motorcycle helmet use in the population of Greater Athens, Greece. Inj. Prev. 5 (4), 264–267. Sosin, D.M., Sacks, J.J., 1992. Motorcycle helmet-use laws and head injury prevention. JAMA 267 (12), 1649–1651. Swaddiwudhipong, W., Nguntra, P., Mahasakpan, P., Koonchote, S., Tantriratna, G., 1994. Epidemiologic characteristics of drivers, vehicles, pedestrians and road environments involved in road traffic injuries in rural Thailand. Southeast Asian J. Trop. Med. Public Health 25 (1), 37–44. Valent, F., Schiava, F., Savonitto, C., Gallo, T., Brusaferro, S., Barbone, F., 2002. Risk factors for fatal road traffic accidents in Udine, Italy. Accid. Anal. Prev. 34 (1), 71–84. Williams, A.F., 2003. Teenage drivers: patterns of risk. J. Safety Res. 34, 5–15.