All the interviews were conducted face-to-face by trained research assistants and lasted approximately 40 min. A parking stage is a well-organized, precise location that registers and allows commercial motorcycle riders to park and wait for passengers. To avoid recall bias, we did not consider cases that could not be invited and interviewed within 14 days after the crash due to poor health conditions or other reasons. Controls were commercial motorcycle riders who reported no history of RTI that led to hospital attendance within the past six months. Recruitment of controls was done at 90 purposefully selected parking stages between December 2018 to March 2019. The parking stages were selected across the five municipalities of Dar es Salaam City Council and included parking stages located at the trunk and collector roads. There are variations in the size of parking stages: those located along the trunk roads have a large number of riders compared to those found in collector roads. This sampling approach was used to ensure that controls were representative of commercial motorcycle riders in the city. At the parking stages, all riders were approached, informed about the study and consented to participate in the study. We recruited about 20% of the riders at each parking stage. We assigned numbers  on a piece of paper and asked riders to choose a number. Compensation of a 1$ voucher was given to riders who participated. Of 413 eligible controls, 13  declined to participate for various reasons, ending up with a total of 400 controls. In total, our study population included 564 commercial motorcycle riders operating within the city of Dar es Salaam, aged 18 years and above . The controls were interviewed face-to-face by trained research assistants. The interviews took place at parking stages and lasted approximately 40 min. The sample size was calculated using OpenEpi version 3.0 statistical software for unmatched case-control study . We aimed to detect an OR of at least 1.8 for being a case, with significant level , at 95% confidence interval, power of 80%, and a ratio of one case to two controls, assuming the prevalence of alcohol consumption among commercial motorcycle riders to be the same  as that of males in the general population .

A minimum sample size of n = 495  was required. After adding 10% non-response rate, the total of 550 sample size was required.The Alcohol Use Identification Test  was used to assess alcohol consumption. The AUDIT is a screening tool developed by the World Health Organization  for assessing alcohol consumption,cannabis grow supplies drinking behaviour and alcohol-related problems . The first three questions in the AUDIT questionnaire are used to measure the frequency of alcohol consumption, the number of standard drinks containing alcohol on a typical day when drinking, and the frequency of heavy drinking. The next three questions covered: symptoms of alcohol dependence, impaired control due to overdrinking, increased salience due to drinking, and morning drinking. The last four questions focus on: harmful alcohol use, guilt after drinking, blackouts, and alcohol-related injuries associated with drinking. All ten items of AUDIT scores were summed up and classified into four categories: non-drinker , normal consumption , risky drinker  and possible alcohol dependence. However, due to the small number of participants with AUDIT scores of 16 and above, the cut-off points of AUDIT scores were revised into three categories; nondrinker , normal consumption , and risky drinker . Missing values for AUDIT scores were imputed using means scores of individual with an assumption that the values were missing at random on less than two items of the scale. Psychoactive drug use was assessed by asking whether the participants had used any drugs in the past 12 months If they responded yes, there were follow-up question as about which psychoactive drug  they had used. The cases were comparatively younger than the controls, with a mean age of 27.8 and 29.5 years, respectively . More than half  of the cases were single, while 59.3% of the controls were married. The cases had a lower level of education and less driving experience compared to the controls. About 39.0% of the cases and 35.5% of the controls reported never having possessed a motorcycle driving license. Regarding the work-related factors, the cases reported working more hours per day compared to the controls. More than half of the cases reported earning a higher daily income compared to 18% among the controls. Nearly 82% of cases reported high risky driving behaviour compared to controls . This study adds to the existing evidence that there is an increased risk of RTIs among commercial motorcyclist riders who report alcohol consumption in sub-Saharan Africa . Our analysis showed an almost six times increased odds of RTIs among commercial motorcyclist riders who reported risky drinking compared to non-drinkers.

The effect decreased to 2.41 times when adjusted for sociodemographic characteristics, work-related factors and risky driving. The decreased effect of alcohol consumption in the fully adjusted model indicates that some of the effect may be due to other risk factors of RTIs Unlike the increased risk of RTIs among riders with risky drinking behaviour, the odds of RTIs was non-significant among commercial motorcycle riders reporting normal alcohol consumption compared to non-drinkers. Furthermore, our findings showed that alcohol consumption in the past year was not significantly associated with the risk of RTIs. This could probably be explained by the fact that alcohol consumption in the past year is a less sensitive measure as it includes both riders who reported normal alcohol consumption and those who fall into the risky drinker category. Evidence from other studies suggested that alcohol consumption and marijuana use are directly or indirectly associated with other risky driving behaviours. In this study we also noted high prevalence of risky driving behaviour in both cases and controls. As motorcycle use is becoming an increasingly important means of transportation in Tanzania , the high prevalence of risky driving behaviour is particularly concerning. Our study shows that most cases were younger, unmarried, and had a lower level of education. These findings are consistent with other studies that have shown an association with increased motorcycle RTIs and sociodemographic factors, notably young age, male sex, and low education level . Only six out of ten motorcycle riders are reported to have ever possessed motorcycle driver’s license even though possession of license is mandatory in Tanzania .Approximately two-thirds of study participants reported working for>13 h a day, suggesting a higher number of working hours in this group of riders. Our findings are consistent with findings from a previous study on risk factors for RTIs among commercial motorcyclists in Uganda by Tumwesigye et al. ,which showed that alcohol consumption was associated with increased risk of RTIs among commercial motorcycle riders , The measurement we have used,the AUDIT is an indirect measure of alcohol consumption which does not measure the effects of the actual intake of alcohol consumption, i.e. level of intoxication at the time of RTIs; instead, it measures the frequency and amount of alcohol consumption to predict the pattern of alcohol use and disorder and hence its influence on the driving ability, directly or indirectly through other mechanisms. Also, studies have shown that drivers with hazardous alcohol consumption are more likely to drink and drive  Previous studies have reported that alcohol intake negatively affects attention and increases the likelihood of sensory and motor dysfunction such as loss of balance and increase in reaction time .

Since balance and coordination are important for motorcycle riders, both novice and experienced motorcycle riders are more susceptible to the effect of alcohol consumption than other motor vehicle drivers . In addition to the impairment of driving performance, alcohol consumption has been shown to be associated with other risky driving behaviours such as speeding, failure to use a helmet, and not having a motorcycle driving license . Moreover, high alcohol consumption has been reported to be associated with non-compliance with traffic rules such as violating the traffic lights, carrying two or more passengers and using cellphones while driving . Our findings show that, even after considering risky driving behaviour, risky drinking was still associated with higher odds of RTIs, emphasizing the need to develop interventions to reduce risky alcohol consumption among these riders. Another important finding revelead that there was a doubling odds of RTIs among riders who reported marijuana use compared to nonusers. Congruent to the observation obtained in a longitudinal study of a cohort of motor vehicle drivers in New Zealand , which showed that statistically significant association disappeared after controlling for risky driving behaviour, cannabis grow facility driving licensure and driving experience. Studies have suggested that increased risk of RTIs observed with marijuana use might appear to reflect the characteristics of drivers who are often young men and prone to engage in risk-taking behaviour . In addition, the studies which have found an association have observed it among drivers who reported to have driven while under the influence of marijuana but not with merely the use . Cognitive studies have reported a plausible biological mechanism of impaired driving performance among marijuana users . However, there has been inconsistent evidence on whether marijuana use increases the incidence of a crash and subsequent injuries in epidemiological and experimental studies . It has been suggested that the mechanism through which marijuana use does not increase crash risk despite neurophysiologic impairments is that marijuana users tend to overestimate their impairment and consequently employ compensatory strategies . Furthermore, driving and simulation studies have revealed that drivers who use marijuana/cannabis tend to decrease their speed, attempts fewer overtakes, and increase their following distance . Moreover, in studies that reported driving impairment, marijuana use was associated with impairment of automatic driving functions of which drivers could not compensate for, and the risk of which increased with increasing doses . On the other hand, other psychoactive drugs such as opiates, stimulants inhalant, and zopiclone have been associated with a higher risk of RTIs . At both the individual and societal level, alcohol consumption and psychoactive drug use represent a severe threat to road traffic safety . Studies have shown that the effects of alcohol consumption increase in dose-related fashion and are more pronounced with complex integrative functions, which is the opposite pattern from that seen for marijuana use . Combining marijuana with alcohol might eliminate the ability to use coping strategies effectively, leading to impairment at lower doses than when either drug is used alone . Our study showed that it is not uncommon among commercial motorcyclists to consume alcohol and use marijuana; 15% of cases and 7.3% of controls reported using both alcohol and marijuana in the past year. However, we did estimate an effect for this combination as data could not differentiate between combined use of alcohol and marijuana at the same time from using them at separate occasions. A notable limitation of our study was that data were self-reported and subject to social desirability bias. Globally, injured victims tend to underreport socially sensitive issues such as alcohol consumption and psychoactive drug use , and a similar observation was documented in Tanzania .

The frequency of self-reported marijuana use among the cases in our study was lower  than what was observed in a cross-sectional survey among trauma patients at the national hospital in Tanzania. The disparities between the findings of our study and this survey could be attributed to the difference in the measurement used. In the trauma patients survey, the psychoactive drug use, including marijuana use, was measured through urine analysis, which is more reliable as compared to self-reports. Also, self-reported measurements may suffer from specific drawbacks due to the way the respondents choose to reply, the answers may be exaggerated or the respondents may be too embarrassed to report a certain behaviours. To address this bias our study following recommendations from previous studies, in particular by guaranteeing anonymity among respondents, clearly defining their roles and the purpose of the data collection and prefacing of questions . Our study included only the commercial motorcycle riders who could be interviewed within 14 days from the date of the RTIs so as to minimize recall biases.