A Review of statistical methods used for intervention evaluation

Abstract

The adopted United Nations General Assembly through Resolution A/RES/74/299 “Improving global road safety” with the Decade of Action for Road Safety 2021-2030 targets to prevent at least 50% of road deaths and injuries by 2030. The global plan called for a holistic approach to road safety and continued improvements in the vehicles and road, enhancement of laws and law enforcement; and provision of timely, life-saving emergency care for the injured. Various road safety interventions and programs have been implemented worldwide with the aim to reduce fatalities and injuries. The importance of evaluating the impact of intervention through sound statistical approaches is definite. As intervention could be conducted in many ways, so can the methods. By design, randomized control trials hold the gold standard in intervention evaluation. However, there are many circumstances where it is not feasible, and researchers opted for a quasi-experimental approach especially when it involves ethical or financial constraints. This paper reviews three approaches used for intervention evaluation: the difference-in-differences method, segmented regression of interrupted time series, and interventional autoregressive integrated moving average, in the field of road safety. The aim is to review the methods used for intervention evaluation or program effectiveness. The Scopus database and available research reports from World Health Organization and related agencies were used to search for available pieces of literature for the year 2013 onwards.