By Carla Navarrete, Pedro de Paula, Rafael Godoy and Sara Whitehead from the Bloomberg Initiative for Global Road Safety
As any ordinary citizen living in or near a major city, you’ve probably passed by a road crash – if you haven’t personally been involved in one. Police are called to the scene, ambulances arrive, patients may be admitted to a hospital where doctors and nurses are rushed in, putting significant pressure on public health systems. Managing the aftermath of a road crash requires strong coordination between multiple individuals and agencies within a city. Investment in the prevention of crashes can reduce this strain, but before any action can be taken, the current road safety situation must be assessed. Practices like data matching play a critical role in gaining this insight, with the goal to later inform policies and interventions that can reduce crashes and save lives.
In cities across the globe, road trauma most significantly impacts its public health systems. São Paulo, Brazil, recorded 797 deaths and 15,455 injuries due to road crashes last year, as published by CET (Traffic Engineering Company). To handle the many costs and consequences of road crashes, the Health Secretariat – responsible for São Paulo’s second largest municipal budget of over USD $2.4 billion in 2018 – devotes a large portion of its resources to road trauma. In addition to providing care for road crash victims, health services expend considerable resources to gather relevant data, such as ICD, provide hospitalization, and perform medical procedures.
Transportation officials need strong data to understand the bigger picture regarding road crashes, fuel policy proposals and shape interventions aimed at crash prevention. Until recently, lack of data integration between the Health and Mobility Secretariats prevented crash-related medical data from being available. This meant that, while having complementary data, these institutions had not been able to take advantage of each other’s records to get a more complete assessment of road safety conditions in São Paulo.
In 2015 as part of the Bloomberg Initiative for Global Road Safety (BIGRS), of which Vital Strategies is an implementing partner, a team of experts began working in São Paulo to implement evidence-based road safety interventions, with the aim to reduce traffic injuries and deaths. As a result, the BIGRS team, in partnership with the Health and Mobility and Transportation Secretariats, identified the need for better data integration as a pathway to saving lives. And in early 2017, an inter-secretarial working group began taking on the technical challenges of integrating large and sensitive databases between the two institutions city-wide to bridge crucial gaps between public health knowledge and action.
One major challenge the group is addressing through this work is direct record linkage – the technical solution to the data integration issue. Due to the nature of those datasets, and how they were initially designed, constructed, filled, and completed for specific purposes, standard identifiers – like names and/or ID numbers – aren’t always available at the time of a road crash. Without standard identifiers, direct – or deterministic – record linkage cannot be done in large datasets. To solve this issue, the BIGRS team proposed a method called probabilistic matching, in which several field values – such as crash/admission date or date of birth/death – are compared between records. The fields are assigned weight based on how closely their two values matches, and then a formula generates the likelihood of a match between two records. This allows those records to be registered as part of the same occurrence (road crash – emergency care – hospitalization – death), and ultimately produce a more complete assessment of road injuries and help design more efficient policy.
Data integration has been undertaken in several countries on both national and municipal levels. However, this is the first time a method of this scale has been used in Latin America. This project has also been a goal at the State’s capital, but in a city as massive as São Paulo (with a population over 12 million), merging these databases is a difficult undertaking.
“We have been trying to implement this work since 2007,” said Lucília Nunes, psychologist from the Surveillance Department of the Health Secretariat. “This type of matching is difficult anywhere in the world, not only here.”
Now that the method is developed and the matching is complete using test crash data, the group will apply this method to current data and begin analyzing the results – with prevention policies in mind.
“To gather data from different sources is a longstanding wish at CET,” said Heloísa Martins, Road Safety Manager at CET. “We now see, with this methodology, that this is possible. The database integration will provide an incredible leap in information quality, besides opening many doors in terms of public policies against road crashes.”
Once completed, the merged data will help provide a better understanding of the real human and economic costs of road crashes in the city, and the interventions needed to mitigate them. Improved data integration will also allow for a comprehensive assessment of CET’s coverage of road crashes citywide.
“If we use the intelligence and evidence from the enhanced information, we can apply the concept of road crashes being predictable and preventable,” said Cheila Marina de Lima, a consultant at the Ministry of Health. “I hope this work can be replicated to other countries and other cities in Brazil.”
By sharing experiences through the road crash data work in São Paulo, the team expects improvements in public health intelligence addressing road crashes, providing a model that can be applied in other Brazilian and Latin American cities.
 Municipal agency responsible for managing traffic data reports on road traffic injuries and deaths
 ICD, or International Classification of Diseases (ICD) is the standard diagnostic tool for epidemiology, health management and clinical purposes, and it is used to monitor the incidence and prevalence of diseases and other health problems, such as road crash injuries.