The study will be conducted in Kampala which has the highest percentage of road traffic crashes 49% per year. Kampala has both public and private roads with an estimated population of ranging from 1.7-2.9 million. Walking is a dominant mode of transportation in Kampala for children despite the inadequate safe walking and crossing facilities. Kampala has 79 Public Schools under the Kampala Capital City Authority.
Study design: A cluster randomized trial will be conducted to assess the effect of school traffic warden on road user behaviour. The 22 cluster study will comprise public primary Kampala Capital City Authority schools. The unit of randomization will be schools. In a cluster randomized trial, random allocation of schools into the intervention and control group will be done using proper allocation concealment when enrolling schools into the respective study arms.
School traffic warden programme intervention The "School traffic warden programme" is an intervention that aims to reduce the risk of pedestrian road traffic crashes at school crossing points through the presence of school traffic wardens at safe points. The intervention will address multiple risk factors related to behaviour. The assumption is that school traffic wardens situated at safe crossing points will increase yield rates of drivers to pedestrians entering or crossing thereby reducing the risk of pedestrian motor-vehicle collision. For school traffic wardens to be effective, they must be placed at safe locations. Some school sites in Kampala already partially have designated safe crossing points. In this instance, visibility and safety at pedestrian crossings will be enhanced by one trained school traffic warden placed a designated safe crossing point. The school traffic warden will assist pupils in crossing at all times during peak hours i.e. 6:00am-8:00am; 01:00pm-02:00pm; 03:00-07:00pm. I will use the non-teaching support employees of the schools as school traffic wardens. In total, 11 school traffic wardens will under-go a 3 day road safety training prior to the start of the study. A similar follow up training will be conducted after 3 months and these will be motivated with non-monetary incentives e.g. high visibility jackets and a "lollipop" stop sign. I will partner with 2 road safety trainers in the delivery of the training sessions. Both the intervention and control schools will receive road safety manuals and leaflets developed by the Uganda Red Cross Society and the Uganda National Curriculum Development Centre.
Baseline assessments will be done for the covariates to ensure that the schools in the intervention and control arms are comparable. Temporal aspects: The study will be implemented over a 6 months period (i.e. the second and third term school study periods of 2022)
Data collection The study instrument will be piloted in 2 schools and feedback used to improve the questions. A team of 4 well trained and experienced data collectors together with one field supervisor will collect data from 22 schools between June and November 2022. The teams will undergo a 5 day training to introduce them to all aspects of the study. Each crossing point will be observed with no breaks during the 3 peak hours of the day (i.e. 06:30-8:00; 13:00-14:00; 16:00-19:00hrs) when children make their way to and fro school. Each crossing point will have a trained research assistant using a camera to record children's crossing behaviour unobtrusively. The camera will be fixed in an elevated position so as to obtain the view of the crossing location. For each crossing point, all individuals observed will be listed and a random number generator used to randomly select 50 pupils. Data will be abstracted on road user behaviour using mobile technology open-source tool for mobile data collection.
Data management Video recordings will be saved in hard drives and upon completion of abstraction, data will be submitted to a secure cloud aggregate server. A daily download of all data from the server will be done followed by data cleaning. As the outcome which is the proportion of road user behaviour is binary, then the generalised regression model will be applied. Analysis will be done using the multilevel model approach.
