Value of multiple correspondence analysis (MCA) to build road traffic crash typologies in low-income countries in Latin America

Abstract

Background and aim: There is a need to identify statistical tools that are both valid and feasible to be implemented in low-resource settings, focusing on identifying individual and event typologies of those at most risk of suffering a road traffic crash (RTC). Multiple correspondence analysis (MCA) has been previously used to build such typologies in other regions. Here, we aimed to explore the typology of both the people involved in the accidents and the events. Method: We selected as a case study the city of Loja, located in southern Ecuador, and conducted a web bases survey. A total of 754 drivers agreed to participate, of whom 41.5% reported having been involved in a traffic accident at least once in life. MCA was used to build typology (profile) of people who have suffered an RTC, and typology of RTC events. Results: Typology of drivers who reported participation or involvement in an RTC was characterized by a predominance of people between 25 and 40 years of age, who drive mainly automobiles (cars) and experience frequent sensations of distraction, and frequently use a mobile phone when driving. Additionally, MCA indicated two distinctive typologies of RTC events. One is characterized by collision vehicle to vehicle, with behavioral factors as a main group of causes, and RTC occurring on low-speed limit roads during the afternoon. The second typology is characterized by collision vehicle-surrounding, occurring on medium speed limit roads during the evening or late evening. Conclusion: MCA suggested the main determinants of RTC are modifiable, mainly related to behavioral factors. MCA is an easy technique to perform, does not require highly trained personnel, and is compatible with information commonly collected in community-based surveys, making MCA a valuable tool for accident research in low-income countries. Keywords: Road traffic crash, typology, multiple correspondence analysis