DECEMBER 2021 1 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS MOBILITY AND TRANSPORT CONNECTIVITY SERIES GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS 2 MOBILITY AND TRANSPORT CONNECTIVITY SERIES © 2021 The World Bank 1818 H Street NW, Washington DC 20433 Telephone: 202-473-1000; Internet: www.worldbank.org Some rights reserved This work is a product of the staff of The World Bank. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions This work is available under the Creative Commons Attribution 3.0 IGO license (CC BY 3.0 IGO) http://cre- ativecommons.org/licenses/by/3.0/igo. Under the Creative Commons Attribution license, you are free to copy, distribute, transmit, and adapt this work, including for commercial purposes, under the following conditions: Attribution—Please cite the work as follows: “Martensen H., G. Duchamp, V. Feypell, V. I. Raffo, F. A. Burlacu, B. Turner, and M. Paala. 2021. Guidelines for Conducting Road Safety Data Reviews. Washington, DC: World Bank.” License: Creative Commons Attribution CC BY 3.0 IGO Translations—If you create a translation of this work, please add the following disclaimer along with the attribution: This translation was not created by The World Bank and should not be considered an official World Bank translation. The World Bank shall not be liable for any content or error in this translation. Adaptations—If you create an adaptation of this work, please add the following disclaimer along with the attribution: This is an adaptation of an original work by The World Bank. Views and opinions expressed in the adaptation are the sole responsibility of the author or authors of the adaptation and are not endorsed by The World Bank. Third-party content—The World Bank does not necessarily own each component of the content contained within the work. The World Bank therefore does not warrant that the use of any third-party-owned indi- vidual component or part contained in the work will not infringe on the rights of those third parties. The risk of claims resulting from such infringement rests solely with you. If you wish to re-use a component of the work, it is your responsibility to determine whether permission is needed for that re-use and to obtain permission from the copyright owner. Examples of components can include, but are not limited to, tables, figures, or images. All queries on rights and licenses should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; e-mail: pubrights@worldbank.org. Contents Acknowledgments...........................................................................................................................................................................................5 Executive Summary.........................................................................................................................................................................................6 1. Introduction: Background and Objectives....................................................................................................................................8 2. Road Safety Data.................................................................................................................................................................................... 11 2.1. Crash and Casualty Data...............................................................................................................................................15 2.2. Mobility Data: Exposure to Risk ...................................................................................................................................29 2.3. Safety Performance Indicators ....................................................................................................................................31 2.4. Data on Road Safety Interventions..............................................................................................................................42 2.5. Reports and Data from Regional Road Safety Observatories...................................................................................43 3. Preparing for the Road Safety Data Review..............................................................................................................................47 3.1. Scope of the Review.......................................................................................................................................................50 3.2. Preparations by the Host Team....................................................................................................................................51 3.3. Preparations by the Review Team................................................................................................................................56 4. Country Visit: Stakeholders to Meet.............................................................................................................................................62 4.1. Ministries and Government Departments..................................................................................................................63 4.2. Police................................................................................................................................................................................64 4.3. Hospitals..........................................................................................................................................................................65 4.4. Statistics Office...............................................................................................................................................................65 4.5. Coroners..........................................................................................................................................................................65 4.6. Insurance Companies....................................................................................................................................................65 4.7. Road Safety Advocacy Groups and Journalists...........................................................................................................66 4.8. Research and Academia................................................................................................................................................66 5. Interviews: Topics to Address..........................................................................................................................................................67 5.1. Organization of Crash Data Collection by the Police.................................................................................................69 5.2. Surveillance of Road Traffic Injuries by Hospitals......................................................................................................71 5.3. Storage, Integration, and Quality Control...................................................................................................................72 5.4. Other Road Safety Data.................................................................................................................................................73 5.5. Capacity...........................................................................................................................................................................76 6. The Review Report.................................................................................................................................................................................78 7. Conclusions..............................................................................................................................................................................................82 Annex A. Sample Monitoring and Evaluation Checklist................................................................................................................ 84 Annex B. Interview Question Sets............................................................................................................................................................85 Annex C. Methodological Issues for Collecting Travel and SPI Data.......................................................................................92 Annex D. Examples of Database Structure.........................................................................................................................................96 Annex E. Sample Terms of Reference....................................................................................................................................................99 FIGURES Figure 2.1. Policy-Making Cycle for Road Safety..................................................................................................................12 Figure 2.2. Overview of Road Safety Data and Their Function for Evidence-Based Management...............................13 Figure 2.3. Aspects of a Database..........................................................................................................................................20 Figure 2.4. Screenshots of the DRIVER Map in Manila, Philippines...................................................................................22 Figure 2.5. Screenshot of Single Record Identification in DRIVER.....................................................................................23 Figure 2.6. Complete Dataflow for the Extract Transform and Load Process.................................................................27 Figure 2.7. The International Road Assessment Programme’s Star Rating.....................................................................33 Figure 2.8. Intervention Indicators, Main Fields and Examples.........................................................................................42 Figure 3.1. Situation Assessment: Comparison between the Official Procedure (Declarations), the Actual Practice (Facts), and the Internationally Recommended Practices (International References)........48 Figure 3.2. Preparatory Activities for Road Safety Data Reviews......................................................................................60 Figure D.1. Crash Database Structures.................................................................................................................................96 TABLES Table 2.1. Minimum Set of Crash Variables..........................................................................................................................17 Table 2.2. WHO Global Road Safety Performance Targets.................................................................................................31 Table 3.1. List of Available Road Safety Data.......................................................................................................................52 Acknowledgments This guide was prepared by the World Bank and the International Transport Forum (ITF), as part of a joint initiative to improve road safety data collection and analysis methods and processes in low- and middle-income countries, led by Veronica I. Raffo (World Bank) and Veronique Feypell (ITF). The document team included Heike Martensen (Vias Institute), Gilles Duchamp (Cerema), Veronique Feypell, Veronica I. Raffo, Alina F. Burlacu, Blair Turner, and Mirick Paala. The team gratefully acknowledges the additional valuable inputs received from Wendy Weijermars, Pilar Zori, Henk Stipdonk, and Fred Wegman. The guide was peer reviewed by Christopher J. De Serio (Senior Transport Specialist, World Bank), Juan Miguel Velasquez (Transport Specialist, World Bank GRSF), Elena Lungu (Transport Specialist, World Bank), who provided helpful recommendations. Maria Segui-Gomez (Consultant, Fédération Internationale de l’Automobile) contributed with further useful comments. This publication was produced with funding support from UK AID through the World Bank’s Global Road Safety Facility (GRSF). Executive Summary 7 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS Accurate road safety data are needed to under- One way to help countries in understanding stand road safety challenges in a country, design possible sources of under-reporting and misre- effective road safety policies, set appropriate porting, and therefore improve their road safety targets, and monitor progress. However, many management, is to conduct a review of their road countries are still in the infancy of accurate data col- safety data. The scope of this review would be to lection, analysis, and use (including related crash data identify weaknesses (and also strengths) of their systems), and not necessarily aware of weaknesses in current process and propose possible measures for their system, which could prevent them from making improvement. An efficient way to conduct this review, sound analysis of their road safety problems. as experienced by the World Bank and the Interna- tional Road Traffic Data and Analysis Group (IRTAD), Effective road safety management requires a permanent working group within the International a broad range of data, such as outcome data Transport Forum (ITF), is to visit a country, meet with (observed crashes and injuries), data on the safety the people involved at different stages in the process performance of the road-traffic system, and data on of road safety data collection, analysis, and use, and interventions to improve road safety. For greater assess the efficiency and effectiveness of road safety impact, these data need to be combined with other data management. traffic data, such as distances traveled and the split between different transport modalities; this approach Following a step-by-step process, the current could be used to identify road safety problems and guidelines aim to support road safety data inform policies and strategies to address them. experts undertaking in-country data reviews. This common methodology will help reviewers prepare As an example, for some countries, an important for their visit, structure their interviews and observa- gap exists between official road death statistics tions, and report on findings and recommendations and estimates by the World Health Organiza- for improvements. At the same time, these guide- tion (WHO), an issue that has been thoroughly lines will help harmonize road safety data reviews debated in various forums, especially in regional across countries and regions, and the results of these road safety observatories. Under-reporting in-country reviews will support the work of regional and misreporting are due to a variety of reasons. road safety observatories. Regional road safety observatories, for example in Latin America, Africa, Europe, and Asia-Pacific, are trying to address this gap and have been generating much-needed interest in the importance of good crash data systems. Several countries have already expressed interest in benefiting from the support of the international community to review and enhance their existing road safety data systems. 1. Introduction: Background and Objectives 9 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS In many countries around the world, deficiencies policy makers on the importance of including in data or data quality impair evidence-based road core road safety indicators is to undertake a safety policy making. While many countries collect road safety data review. Such reviews are typically road safety data, the collection is not necessarily conducted during a visit to the country by a team comprehensive. Further, many countries could be of independent road safety data experts. The team unaware of data gaps in their system, which pre- should have an excellent knowledge of road safety vents them from soundly analyzing their road safety data and offer sound perspectives on international problems. Therefore, road safety data definitions good practice. Review teams should also have a solid and collection methods must converge into standard understanding of the local context, which often leads international criteria, thus allowing for compari- to heterogenous teams with a mix of local and inter- sons in space—across countries—and in time. This national experts. is the raison d’être of regional road safety observa- tories, which have been developed, for example, in The following guidelines are designed to support Latin America (OISEVI), Africa (ARSO), and Asia-Pa- reviewers in the assessment of road safety data cific (APRSO). They present an opportunity for joint collection; the complete range of safety data should regional efforts to improve, in a harmonized way, be considered. This task can be complicated because road safety data collection and analysis. Regional collection of road safety data is often not achieved road safety observatories promote the adoption of by activities dedicated to this purpose, but rather a common set of road safety indicators based on through piggybacks on other sources. For example, common definitions and serve as an avenue to assist activity reports from police or hospitals are used to countries in improving the management of their provide material for legal or medical purposes. The crash data systems. routines involved frequently have a long history in which gathering reliable and complete statistics has Road safety data are not just about crash data (or had secondary priority, at best. The various actors outcome data—the observed crashes and injuries), involved reflect the complex structure of a country’s but also about the safety performance of the road judicial and executive system, which, generally, are traffic system, and about interventions to improve not coordinated. Consequently, any review of the data road safety. These data are best used when com- collection process requires some “detective work.” bined with other information, such as traffic vol- umes and distances traveled, or split between dif- The guidelines shared in this report address vari- ferent transport modalities. For an evidence-based ous aspects of the whole data collection process, approach to the management of road safety, these such as road crash investigations, reporting and data can be used by policy makers, traffic engineers, registration, checking completeness and consis- police, the health sector, the research community, tency, storage, and accessibility—as well as anal- insurance companies, prosecutors, vehicle manufac- ysis and use. By considering the experience from turers, and others. earlier data-review missions, the guidelines indicate typical problems for each of these aspects, along with One way to help raise the quality of the most examples and questions useful to the review team essential data and create awareness among visiting a country. This guide briefly summarizes the 10 MOBILITY AND TRANSPORT CONNECTIVITY SERIES importance of road safety data and the international of questions based on the materials provided by standards for the most important indicators and the country. The guidelines address various aspects describes typical problems in their application. Con- of the entire data collection process, such as crash sequently, the guidelines can also be of wider general investigations, reporting and registration, check- interest and serve as a useful self-evaluation tool for ing completeness and consistency, storage, and road observatories. accessibility. The guidelines will help harmonize road safety A data review should consider three main activities, data reviews. Apart from questions to ask and namely potential problems to investigate, they suggest how to describe the strengths and weaknesses of current a. preparatory activities, such as conduct- processes and how to formulate recommendations ing research and developing preliminary with respect to organization, methods, training, and assessments; communication. While much of this information is available in other documents, these guidelines b. conducting interviews; and present in a compact document the information most important to consider in a data review, and are c. drafting the report. intended to help the review team develop its analysis. More concretely, the guidelines describe the neces- The next chapter opens with a general description sary preparations of both the reviewer and the coun- of road safety data, its role in decision making, and try under review. They suggest people and organi- other important aspects worth considering. zations to meet during the visit and the preparation 2. Road Safety Data 12 MOBILITY AND TRANSPORT CONNECTIVITY SERIES Road safety data are important for evidence-based The implementation of the countermeasures must be decision making. This chapter sets a foundation monitored and their impacts evaluated. The process by explaining why road safety data are important. is then repeated in a cyclical way, as shown in figure In addition, international standards and issues to 2.1 (WHO 2010; PIARC 2019; ITF 2016). consider for different types of road safety data are discussed. In particular, crash and casualty data, To better understand risks and assist in the monitor- mobility data, safety performance indicators, and ing of progress in road safety, crash data are import- intervention indicators are addressed. ant, but alone are not enough; they must be com- bined with other types of data. An evidence-based approach to road safety manage- ment starts with the definition of the problem: risk When setting priorities, the number of casualties factors must be identified and prioritized for treat- must relate to the share of this group in traffic. ment. On this basis, the actions and targets need to As an example, to address pedestrian casualties it is be defined during the initial stage of the diagnostic. important to know whether the numbers are high because many people walk, or because walking is Figure 2.1. Policy-Making Cycle for Road Safety 1 Define problems 5 Monitor performance 2 Identify risk factors and priorities 4 Implement strategy 3 Formulate strategy and set targets Source: Adapted from WHO 2010. 13 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS particularly unsafe (or both). For this, one needs If a modifiable risk factor has been selected for exposure data specifically on the number of users per treatment, it should be measured. These road travel mode. Moreover, it is important to distinguish safety measurements are known as road safety between modifiable and nonmodifiable risk factors. performance indicators (SPIs) and examples include As an example, the age-structure of a country’s pop- the proportion of car occupants who do (not) wear ulation cannot be changed. So, statistics on these a seatbelt or the proportion of drivers who drive issues are most useful if they also indicate specific sober. For target-setting and monitoring it is often group concerns that can be modified (for example, better to consider the SPI addressed directly rather statistics might look at the share of crashes involving than evaluating the number of casualties related to children crossing the road). it. This is especially important if the risk factor cannot be reliably identified in the crash data (for instance, Risk factors should be selected for treatment if whether the seatbelt was worn). Note, however, that they are modifiable. For example, the population SPIs are only valid tools if their link with the desired structure is given, and road safety policy should focus outcome (that is, reducing casualties) has been pre- on the most promising aspects to change, such as viously proven. Figure 2.2 provides a summary of all the usage of seat belts or the layout of roads. road safety data types and how they are linked to effective road safety management (PIARC 2019). Figure 2.2. Overview of Road Safety Data and Their Function for Evidence-Based Management Results Costs: Medical costs, material and intervention costs, productivity loses, traffic jams (lost time), loss of life/quality of life Outcome indicators: Social cost Crashes, injuries, deaths (combined with exposure data) Final outcomes Safety performance indicators: Speed, alcohol, restraints, helmets, road Intermediate outcomes infrastructure, vehicle safety, trauma Process/implementation indicators: Outputs Road safety policies, plans, programs Interventions Road Network Planning, Entry and exit of Recovery and design, opera- vehicles and rehabilitation of tion and use drivers crash victims Institutional Results Focus Management Functions oc ur d n a g tra led nd ion ion all eso an rin tio at ce at nd ns ge o a ion ion r ina ito fe lat r ng kn &D ot om on gis rd i R nd alu o M Le Pr Fu Co ev Source: PIARC 2019. 14 MOBILITY AND TRANSPORT CONNECTIVITY SERIES The desired results or outcomes of road safety The focus should be on using the available data— management should be expressed as goals and even if they are not perfect. The recommendations for targets at different levels of the data pyramid: (a) collecting new data should be based on careful con- Interventions that have been implemented; (b) inter- sideration of the benefits as well as associated costs. mediate outcomes; and (c) final outcomes. Moreover, contextual data, such as the road network and the Road safety data include outcome data (the volume and type of traffic, are important for prioriti- recorded crashes and injuries), data on the safety zation. These latter factors do not only inform policy performance of the road-traffic system, and data making, but could also be addressed by it—such as a on interventions to improve road safety (see policy on accessible and safe cycling and pedestrian figure 2.2). To interpret crash data, traffic and other infrastructure to influence shifts from cars to more contextual data are also important. In a good man- sustainable modes of transport. agement system, these data are used to analyze risks, plan interventions, evaluate efficacy, and guide As an example, for this process, a program to development by setting targets for outcome and address the injury risk of motorcyclists and increase performance indicators in monitoring their progress. the usage of helmets would be evaluated in terms of To allow for this, not only is the quality of the data numbers of motorcyclists checked (program out- important, but also their accessibility and use. All put: interventions), percentage of motorcyclists with these aspects should be addressed in a data review. helmets (SPI, intermediate outcome), the severity of motorcycle crashes (final outcome), and (if possible) Road safety data do not only exist at the local and the percentage of motorcycle casualties who did and national levels, but also at the regional and global did not wear their helmet. The size of the motorcycle levels, including those from regional road safety fleet would be included in an analysis to correct for observatories. A major aim of regional observato- other trends (for example, an increase in motorcycle ries is to address issues on data by assessing how traffic). For more background information on road national statistics and database information on road safety data and their use to reduce road traffic casu- crashes can be improved by reporting better quality alties, see for example, Papadimitriou and Yannis information. This, in turn, facilitates more appropriate (2018) and Wegman (2016). solutions and results in better cross-country compar- ative data between countries. Looking at the outputs Note, however, the cost to information. All recom- then of regional observatories can also guide and mendations should consider the specific data points further contextualize the data review. actually needed to take decisions. Only those data that will change decisions are worth the investment The remainder of this chapter lists different types in data collection. A huge amount of knowledge of road safety data and discusses international best about road safety interventions can be formulated practices and minimum standards for each type. and implemented without gathering any new data. 15 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS 2.1. Crash and Casualty Data Crash and casualty data should give the full pic- The data should be complete, of good quality, and ture of road risk. Crucially, completeness depends collected uniformly throughout the country. The on the notification of crashes occurring, the capac- following aspects should be considered to ensure this ity to attend the crash scenes, and the registration is the case. and transmission of data. Usability of the data also depends on the variables registered, correct registra- COMPLETENESS AND NOTIFICATION tion of the crash location, storage, and accessibility In many countries not all crashes are reported. as well as linking different data sources to check for This is particularly problematic if some types of completeness and augment the data. Data on the crashes are much less likely to be reported than most severe crashes (fatal and serious injury) are the others, because it creates a skewed picture of the most important. situation. A typical example concerns crashes with no motor vehicle involved. For instance, a study on The count of road traffic crashes and their casu- hospitalized cyclists in 17 countries showed less alties should include every crash of the specific than one-third of cyclists crashed with a motor vehi- severity being captured (for instance, all fatal crashes cle and among the others (cyclists who fell, hit an or all injury crashes). Whether the crash is reported object, crashed with another cyclist or with a pedes- should not depend on the types of road users trian) less than 20 percent reported their crashes to involved, or where and when the crash took place. the police (Shinar et al. 2018). Often insurance pay- ments are the main reason to report crashes to the Data collection should be prioritized according to police; crashes where no payment can be collected the severity of the crashes. Data on fatal crashes would be reported less often. Additionally, citizens have the highest priority, followed by data on severe who lack trust in the police are more likely not to file injuries. Data on minor crashes or property damage reports. In addition, the lack of staffing and resources only crashes—while important—have lower priority. can also impede police and emergency response, especially in rural or far-flung areas. To investigate To be in line with international definitions, fatalities possible sources of under-reporting, it is important should include deaths up to 30 days after the crash. to know how the police are notified of crashes and Severe injuries are typically defined as resulting in a if some crashes are typically not reported to them. hospital stay of 24 hours or more (see Azzouzi 2019; Furthermore, in certain contexts, culture, and social CARE Team 2018, 133). Slight injuries are defined as customs influence the reporting of crashes—specif- those needing medical treatment (Eurostat, ITF, and ically of crash fatalities. In Vietnam, for example, a UNECE 2019). This also requires coordination among person who is about to pass away is usually brought government agencies, particularly of the police and home from the hospital. This, in turn, prevents hospi- the health department, to standardize definitions and tals from recording complete fatality data. Alongside forms, update crash data with injury data, and possi- internal checks, the numbers should be compared to bly even integrate the health database with the crash other sources to ensure completeness. database. 16 MOBILITY AND TRANSPORT CONNECTIVITY SERIES UNIFORMITY must have enough time at the scene and have the Often, death and injury registration are not technical means—such as sketching the scene, using recorded by the same institution. In such cases, an alcometer to measure blood alcohol levels, among hospitals, police, and the civil registry are the usual other factors—to document the crash thoroughly and actors. These different actors must coordinate with register all variables in the crash report form. each other to standardize and align data reporting requirements and data sharing arrangements to Ideally, causation and aggravation factors should validate respective agency data. This does not only be recorded, such as speeding, driving under the concern different types of institutions (hospitals and influence, seatbelt use, and other violations. Impor- police, for instance), but also different levels and enti- tantly, the review should investigate whether the ties within the same institution. For example, some police in fact have the requisite tools and resources countries have different police units for different for identifying such factors. administrative or jurisdictional regions of the country. In other instances, the police from the local, regional, The severity of the victims’ injuries should be and national levels will not be able to store and share reported. Officers at the scene can use several tech- their data in a single repository. The recording sys- niques to determine injury severity. One of these is to tems of different entities should be sufficiently simi- give subjective assessment, but this does not nec- lar. All institutions should apply the same definitions essarily agree with the medical diagnosis. The initial related to crash injuries. assessment of the officer at the scene should there- fore be updated later based on medical records. These ATTENDANCE AT THE CRASH SCENE updates, however, do not always happen. Special To give a complete picture of road crashes in a attention should be paid to victims who die in hospital. country, the police should ideally attend every Their inclusion in the fatality count must be ensured. crash scene—at least for crashes resulting in serious or fatal injuries. This should apply even to In some countries the police do not investigate challenging situations (for example, remote areas, all crashes. Where possible these events should still night-time, or multiple crashes happening at the be documented, together with recorded information same time). Otherwise, crash types somehow difficult about any crash characteristics, and how that infor- to attend to are structurally under-reported. Officers mation was gathered. 17 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS VARIABLES REPORTED of variables derived from the common Accident Data In addition to the data police collect for prose- Set (CADaS) of the European Commission (Azzouzi cution purposes, a minimal set of data must be 2019; CARE Team 2018). In its minimal form (miniCA- collected that gives information about the location, DaS), the protocol indicates a set of 28 variables that the infrastructure, road users, and vehicles involved should be recorded for every crash (see table 2.1). For as well as variables characterizing the maneuvers information about the possible values that should be and the consequences of the crash. The African foreseen for these variables, see for example CARE Road Safety Observatory (ARSO) has adopted a set Team (2018). Table 2.1. Minimum Set of Crash Variables Crash Traffic unit Person Crash ID Person ID Date Traffic unit type (e.g., pedestrian, Date of birth cyclist, passenger car) Time Special function of vehicle Gender Weather conditions Registration year Road user type (pedestrian, driver, passenger) Lighting conditions (daylight, dark, with/without Country of registration (e.g., Injury severity (slight, more than 24 hours in lighting, dusk/dawn) foreign, national) hospital, fatal) Crash type (e.g., with pedestrian, single, two Vehicle maneuver (e.g., turning, Alcohol test (not tested, not applicable, positive, vehicles turning, two vehicles NO turning) overtaking, etc.) negative, unknown) Location: X coordinate (latitude) and Y coordinate Drug use (longitude) Road type (e.g., motorway, expressway, national Safety equipment road, local road) Section type (e.g., bridge, tunnel, bend, Nationality (national, foreigner—possibly by gradient, straight) relevant country grouping) Junction type (not at junction, crossroad, MAIS injury severity (Maximum (M) Abbreviated roundabout) Injury Scale (AIS) rated from AIS 0 – no injury, to AIS 6 – death) (AAAM 2016). Speed limit Surface conditions (dry, snow/ice, wet, slippery) Crash severity Source: Adapted from AAAM 2016; Azzouzi 2019; and CARE Team 2018. 18 MOBILITY AND TRANSPORT CONNECTIVITY SERIES Additional variables in other minimum datasets the other minimum requirements) enables the use include, for example, curvature, division, surface of modern tools, such as geographic information (sealed or not), junction control and its function- system, or GIS-based crash data processing for the ing for the road and license type, airbag state, and automatic pre-calculation of black spots. Without reli- seating position for the persons involved (Aus- able knowledge of crash locations, opportunities for troads 1997). In addition, it is also helpful to include solving local deficiencies remain limited. a description of the reporting agent and a diagram showing the crash in relation to local road features, Missing or inaccurate reporting of road crash the direction of travel of different vehicles and road locations is one of the most pressing challenges users, and what happened in the crash. in road safety analysis aimed at improving safer road systems. The availability of an exact road loca- Some types of data elements can be derived by tion reference method is therefore seen as the most integrating the crash database with other data- essential element of a traffic information system: base systems. This integration can simplify reporting and validate the information collected at the scene. • Road authorities need to pinpoint crash loca- For example, data regarding the person (such as date tions accurately to tackle hazardous locations of birth and nationality) can be derived if the crash effectively; police need this to efficiently allocate database is integrated with the license register or enforcement resources, and hospitals need this social security data. Vehicle information can be added to optimize their emergency response. Missing via the vehicle registration database. Road infrastruc- or inaccurate location of crashes might not only ture data (including junction type and road type) preclude the identification of the worst sites, but can be derived if crash data contain georeferenced could also hinder the ability to evaluate the effec- information and are connected to a geospatial road tiveness of any countermeasures. database (see the “Crash location” section, below). The Asia-Pacific Road Safety Observatory (APRSO) • The location reference system can provide has recommended specific core crash data elements a link between various files (crashes, traffic to be collected at the crash scene and data elements data, and roadway inventory). To successfully that could be derived from other sources, such as an merge these files, the location reference methods inventory of roads and bridges, licensing and vehicle used in each of them should be identical or, at registration database systems, and injury surveillance least, compatible. systems. Each road crash relates to a specific location in CRASH LOCATION the road network. The two major options for iden- The location of a crash is a critical data point and tifying crash locations are via GPS (global positioning one of the minimum requirements in effective system) or through road-based identification: data collection and use for identifying potential road safety interventions. The availability of crash • Use of GPS: GPS is the fastest and most exact way locations on an electronic database (together with of obtaining reliable data about crash locations 19 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS everywhere on earth. Localization can be per- REGISTRATION AND TRANSMISSION formed at the road crash location—mobile GPS The data should be registered in a common sys- devices have sufficient accuracy to obtain location tem and transferred from local police stations data (geographic coordinates). Even if GPS coor- to a central data repository where all crash data dinates cannot be obtained at the crash location are consolidated. All possible problems with this (due to malfunction of the GPS-device, for exam- procedure should be identified. Ideally, crashes ple), the coordinates could be determined ex post should be registered by the person who attended by using available data on a national geographic the scene; however, this is not always the case. If the information system (GIS) platform or from the crash is registered by someone else, this must be internet. done without any significant loss of information. Data input should not be too time consuming and done • Road-based identification: Depending on loca- preferably with a user friendly interface to ensure tion, two different systems could be used for this: completeness and avoid errors. Optimizing the form used (such as electronic or paper) can bring major o Stationing in countryside areas: This is the tra- improvement. The variables recorded should at least ditional and most commonly used method for include those listed in the “Variables reported” sec- road location identification in the countryside. tion earlier in the chapter. Further variables can be The road identification code or number and the of additional value, but only if all fields have been stationing data (mileage of the road) describe reliably completed. each road location. Each road has an origi- nal station (zero point) and the distance from An investigation should be undertaken of the this point defines each location. Distances are subsequent handling of registered cases and how normally indicated by kilometer post markers, cases are collected at national level. It is common available during crash data recording. A disad- that data are collected then kept at station level—in vantage of this system is its inadaptability to a filing cabinet or in a pile, which often results in a infrastructural changes that result in a change major barrier to improving data systems. of the road length, along with the difficulty of making a unique link with XY coordinates. Special attention should be paid to the nature Another difficulty is that in some countries sta- (paper mail, email, cloud) and frequency (daily to tion points do not exist, or do not exist every- annually) of the links between actors involved where depending on road status. in the data chain, and to the possible data nodes where data are gathered for verification, correc- o Address in urban areas: A street address is tion, or processing. The time period for the data to commonly used in urban areas. However, as reach the crash database following a crash should be for station points, street addresses do not exist documented. everywhere and in every country. 20 MOBILITY AND TRANSPORT CONNECTIVITY SERIES DATA STORAGE: THE CRASH DATABASE A database is built from a process comprising four Whatever the database system in use, the main steps: reviewer must always keep in mind the follow- ing paradigm: Check that all the data stored are 1. Analysis of the existing situation and needs needed, and check that all the data needed are stored. A database is a voluminous set of varied data, 2. Creation of a series of conceptual models to rep- stored in centralized or distributed digital media to resent all the important aspects of the problem minimize redundancy and to link the different types of data. The database serves the needs of one or 3. Translation of the conceptual models into a logical more organizations, so that (potentially different) model and standardization of this logical model stakeholders can feed the data, correct, modify, and enrich them; and query the database (see figure 2.3). 4. Implementation of a database in a data manage- ment system, based on the logical model and optimization Figure 2.3. Aspects of a Database Data Database Stakeholders Store (centralized or Feed Voluminous distributed) Minimize Correct, redundancy modify, enrich Structured Link Query Source: Original figure produced for this publication. 21 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS A database management system is software aiming At this stage, several roles can be identified: to manage the data with an adapted language, in which: • Contributors: Collect and provide required data for the database (such as hospitals or police); con- • Information is stored and organized as objects or tributors may or may not be editors tables; • Editors: Enter the collected data into the database • Tables are sets of fields describing the subject of the table; and • Database administrators: Manage the database consistency and assign access rights (reading, cor- • Tables are linked by one or more relationships recting, deleting, validating) between fields, some of which are indexed. • Validators: Oversee declaring the data as valid for The data review focuses on each of the different publication. bullet points above, bearing in mind these require- ments are not necessarily linked to a particular tool, At this stage, the official and labelled database, once although it is easier to achieve the objective with an checked and published, must be clearly differentiated online shared and dedicated tool than with individual from the “living” database, which could evolve after Excel or other spreadsheets. the date of official validation with new information that enriches or modifies it. The structure of the database and the tools used to implement it are not unique. Several examples DATA QUERYING AND PRESENTATION can be given worldwide of different choices (see The database should feed into analysis tools. appendix D). Nevertheless, some key points must be Sometimes, these tools are so integrated into the checked: user interface they could appear to be embedded in the database itself. This is, however, not the case. The • A data dictionary must exist, and be known by the operating tools, even if linked to the structure of the actors of the data chain database, are not part of it. The most efficient sys- tems clearly distinguish storage and exploitation. • Concepts like “not applicable,” “unknown,” and Querying the database must allow the combination “missing” must be identified of different variables, such as selecting all crashes of child pedestrians on weekdays, just before school • Every change in the database must be tracked and starts. Obviously, queries on the database must take time-stamped. account of and be consistent with the database struc- ture. As an example, in a database the characteristics 22 MOBILITY AND TRANSPORT CONNECTIVITY SERIES of the road users might be stored, along with the Road Incident Visualization Evaluation and Reporting characteristics of the roads leading to an intersection (DRIVER) system (figure 2.4), a web-based and open- at which a crash took place; however, the two might source platform for geospatial recording and analy- not be unequivocally linked. In that case, statistics sis of road crashes developed by the World Bank in combining the type of road user with certain charac- cooperation with local governments and various data teristics of the roads cannot be produced (or worse, providers, including Mapillary, a street-level imagery they might be produced but not make any sense). platform. DRIVER allows filtering according to sev- eral variables, production of heatmaps and blackspot The querying system should include functions analyses, and zooming into single records (as shown that allow mapping of crash locations and dis- in figure 2.5). play information on areas, routes, or locations in a useful way, such as in a factor matrix or as collision Crashes can be analyzed based on location, facilitat- diagrams. Ideally, the system should also include a ing black-spot analyses and allowing users to relate template for a standard report that can be generated crashes to road characteristics (see also the section by different administrative units, such as communi- discussing “Road user behavior” later in this chapter). ties, provinces, and others. An example is the Data for Figure 2.4. Screenshots of the DRIVER Map in Manila, Philippines Source: Burlacu 2019. 23 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS Figure 2.5. Screenshot of Single Record Identification in DRIVER Source: Burlacu 2019. 24 MOBILITY AND TRANSPORT CONNECTIVITY SERIES Box 2.1 provides an example of how the DRIVER platform is used in Lao PDR. Figure B2.1.1 Screenshot of the DRIVER Platform in Lao PDR Source: Lao PDR DRIVER platform (http://laos.roadsafety.io). To date, a total of more than 2,000 crash records have been mapped and encoded into the DRIVER platform in Lao PDR. More than 100 people, including traffic police, have been trained in using DRIVER. These include police at the local level, the central Department of Traffic Police as well as staff from the Ministry of Public Works and Transport (MPWT) and the National University of Laos. The Lao DRIVER platform has been translated and updated into the Lao language and is currently accessi- ble at http://laos.roadsafety.io (see figure B2.1.1) Currently, the MPWT and the Department of Traf- fic Police, with support from the World Bank, are in the process of scaling up DRIVER at the national level through the signing of legal instruments, standardizing data collection processes and forms, and securing technical resources, such as servers. A preliminary institutional arrangement for DRIVER implementation has been developed for Lao PDR, where the police, hospitals, and other ministries feed data into DRIVER, which can then be accessed by relevant stakeholders (figure B2.1.2) 25 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS Figure B2.1.2. Institutional Arrangement for the DRIVER Platform in Lao PDR Crash data from Injury data from Road infrastructure Other road MOPS hospitals data from safety data MPWT Injury data from hospitals Lao Driver Instance Instructure and Researchers Government Advocates Transport Providers Source: Original figure produced for this publication. SHARING DATA “owners” should ensure the engagement of different The data should be accessible, both in terms of stakeholders—including police, road administration, access rights and in terms of comprehensibility of engineers, insurance organizations, lobby-groups, information. Ideally, the actual data should be acces- and local/regional/national policy makers. They have sible to experts with a good understanding of the to make sure stakeholders can derive the information data structure and include an access tool that is easy they need, for example, by producing dashboards for all to use. Rather than “sit on the data,” the data and maps or by making standard reports available. 26 MOBILITY AND TRANSPORT CONNECTIVITY SERIES CHECKING AND AUGMENTING THE DATA location can be used. If records meet a set number of The crash database could be checked and aug- matches with these secondary identifiers, the two will mented using external data. Thus, the following then be linked or merged. Finally, creating a custom- questions need to be addressed: ized crash identifier, such as a record number, will facilitate the linkage of the two database systems. • Are the data checked against other counts, such as hospital data, vital statistics, and coroners’ Notably, most current database systems are reports? Is the completeness of the different data- developed in open environments that allow easy bases known or estimated? interfacing with external data, which is stored in warehouses using an extract, transform, and load • Is double counting possible or checked? Are (ETL) process. The ETL process allows the validation, plausibility checks routinely conducted? Are the cleansing, transformation, loading, and aggregation results fed back to those who originally record the of data for storage: crashes? n a data warehouse, generally seen as a central- • I • Are the data linked to and enriched by other ized storage space grouping together data from sources, such as hospital data and vehicle different sources, thus making them homoge- registration? neous or belonging to one uniform dataset; or To achieve integration between crash data and • Possibly in specialized partitions of a data ware- a nonspatial dataset (health data, vehicle regis- house (data mart) after validation, transforma- tration, driver licensing) will require matching tion, and cleaning, so as to be available in various of identifiers or data elements. These identifiers forms to assist with decision making. can be (a) unique identifiers, (b) secondary identifi- ers, or (c) customized identifiers. Unique identifiers These are then made available to users for all pur- include data elements specific to a single record, poses: Online Analytical Process (OAP), data mining, such as personal identification numbers and names. cartographic visualization, dashboards, and indica- If using unique identifiers is not possible, secondary tors. The complete dataflow is illustrated in figure 2.6. identifiers such as the gender, date of birth, age, and initials of any victims as well as the crash date and 27 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS Figure 2.6. Complete Dataflow for the Extract Transform and Load Process From To sources publication Databases Online analytical process (OLAP) Validation Data Open data mining Cleansing Transformation Loading Cartographic Aggregation Data warehouse visualization Data marts IT servers Dashboards Indicators Documents Source: Duchamp 2017. However, integrating datasets into a single data- Box 2.2 looks at the development of the National base can produce duplicated records. To address Road Safety Database in the Republic of Serbia. this, a mechanism for duplicate record management must be in place. A dedicated office that checks the overall quality of data, including resolving potential duplicates, must be established. 28 MOBILITY AND TRANSPORT CONNECTIVITY SERIES Box 2.2 National Road Safety Database: Republic of Serbia In 2016, the Serbian crash data collection system was revised, to harmonize with the European Common Accident Data Set (CADaS) and the system of causation factors used in the UK STATS19 accident data collec- tion forms. The system was designed in a close cooperation between local police who collect the data and transport engineers who use the data. Training programs, automatic checking functions, and feedback to the officers who enter the data all help ensure a high level of quality. The crash data files are augmented with a large number of other data files gathered from the Ministry of Internal Affairs, the Road Traffic Safety Agency, and the State Road Agency Public Enterprise, “Roads of Serbia.” Together, the files form the national road safety database of Serbia (see figure B2.2.1). All information is GIS coded, allowing presentation of results as maps. Two tools, each tailored to different users, allow access to the information: 1. The public can access the information via a web tool (http://195.222.99.60/ibbsPublic/) that allows selec- tion of accidents or victims according to many criteria, including the year, severity, vehicle type, road user type, age group, accident type, causation factor, road type, area type, among others. The results can be shown on the map or exported to spreadsheet files. 2. The researchers have a more advanced web-based tool that still allows them to quickly create queries in the database while combining criteria more freely. The advanced tool also includes all variables available in the database. Figure B2.2.1 Flowchart for the National Road Safety Database of Serbia Source: Martensen 2018. 29 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS 2.2. Mobility Data: Exposure to Risk To identify road risks, traffic data should be avail- represent a threat to road safety when motorized able and as detailed as possible. Ideally, these data transport and vulnerable road users are mixed in should be available for each type of road user, includ- traffic. However, as the share of active transport ing nonmotor vehicles and pedestrians. This section modes increases, more space in traffic is often ded- discusses various ways to collect these data, as well icated to them, including pedestrianized road corri- as surrogate data that can be used in the absence of dors and dedicated cycle lanes or paths, which offer data on travel distances. safe ways to travel. Data on traffic volumes are important for the TRAVEL DISTANCES analysis and interpretation of crash data, because The gold standard of crash and casualty analyses they quantify the numbers of at-risk units. Such units is to link them to distances traveled, because this can be persons (population, license holders), vehicles allows the most direct link to countermeasures. (fleet attributes such as class or age), road sections By knowing how people travel (by car, bicycle, on (road lengths), or ideally instances of vehicles or per- foot, along with other modes) and how far, one can sons moving in traffic (distance traveled, trips, min- distinguish differences in road risk from differences utes in traffic). They determine the total societal risk in travel behavior. because some travel modes present more risk than others. For instance, a higher proportion of travel by Ideally, travel distances are estimated with travel motorcycle results in a higher number of casualties. surveys. Traditionally, a representative sample of the population is required to fill in a mobility diary (usu- Risk is estimated by dividing the number of ally for one day) and report on each trip made as well crashes or casualties by a measure of exposure, for as the modes of travel used. The estimation of dis- example, the number of fatalities per 100,000 inhab- tances and travel times by the travelers themselves itants or per billion kilometers traveled. Note that for is very unreliable. As a solution, rather than sampling this reason the risk must always be defined in terms persons willing to fill in a questionnaire, one can sam- of the numerator and the denominator. A good over- ple persons willing to install an application on their view of exposure data for road safety analyses with smartphone to measure travel modes, distances, practical examples for their collection can be found in routes, and travel speeds. Examples are available the SafetyNet project (Yannis et al. 2005). Appendix C from Switzerland (https://ivtmobis.ethz.ch/mobis/ describes methodological considerations for the col- covid19/en/) and the Czech Republic (https://www. lection of exposure measures in more detail. czrso.cz/nub/post/map). The big advantage of sur- vey data is that they include the distances covered by With a growing understanding for the need to walking or by nonmotorized vehicles. Another alter- shift traffic from motorized transport to more native is to work with the data that big data collectors active modes such as walking and cycling, the can provide, although often little is known about the modal share of different forms of transport are representativeness of the underlying sample. In a increasingly important. The modal share holds first investigation conducted in 2021, researchers in important implications for road safety and can Finland concluded relatively large differences exist 30 MOBILITY AND TRANSPORT CONNECTIVITY SERIES between their own estimates of kilometers traveled GDP, the final measure on the list, is the most indirect and those estimated by Google (https://www.google. one, but also the one most readily available in most com/covid19/mobility/).1 countries. Its relation to road safety is complicated, in the sense that richer countries usually have better Traffic counts offer an alternative. Traffic count road safety records (Kopits and Cropper 2005), while systems allow for continuous measurements of traffic at the same time in these same countries, road safety volumes over time. Traditionally, only motor vehicles improves when times are economically hard (ITF have been counted. However, modern video observa- 2015). Nevertheless, all results are based on calcu- tion techniques can also monitor walking and other lations of risks per travel distance. The GDP can be nonmotorized modes of transport. considered a good proxy of economic activity and in consequence mobility (Antoniou and Yannis 2013). SURROGATE EXPOSURE MEASURES If no travel distances are available, surrogate MODAL SHARE measures should be used for the interpretation of If detailed distances per travel mode are unavail- crash and casualty figures. These usually do not dif- able, general estimates described in that chapter ferentiate between modes and are therefore clearly above can be supplemented by indicators for the less preferable to the more detailed data described share of different travel modes. While the detailed above. However, as the number of participants is still travel surveys described above remain the ideal most strongly determined by the amount of motor way to investigate modal share, insights on modal traffic (a pedestrian will rarely cause fatal injuries to share can also be based on other data. Very gen- another road user in a crash), these numbers can eral questions on which modes are used most often still be helpful in understanding the development can provide a rough overview, cost much less than of crash numbers in countries with growing motor- a complete travel survey, and could also be added ization. Moreover, they are often readily available in with other questionnaires. Moreover, many compa- economic statistics. Surrogate measures include the nies maintain data on how their employers travel to following: work. Mobility platforms such as Uber also have data on the number of drivers using motorized vehicles • Fuel consumption as opposed to bicycles. Pedestrian travel is the most difficult to estimate. However, the vehicle rate per • Road length household is a good proxy for that. • Vehicle fleet (from vehicle registration) • Driver population (from license registration) • Gross domestic product (GDP) 1 Google data compared to other traffic volume data sources in Finland. Taken from translated excerpts of an internal ITF/IRTAD memo written in December 2020 concerning telecommuting and COVID-19. 31 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS 2.3. Safety Performance Indicators The transport system consists of road users, their In principle, SPIs should be chosen in relation to behavior, the roads they use (that is, the infra- the safety issues in the country under review. To structure), and their vehicles. All three components determine the relevant intermediate outcomes, the contribute to the safety performance of the system. crash data must be analyzed, and risks identified. For each component, this section discusses the dif- As an example, if a high proportion of pedestrian ferent indicators of safety performance and the main fatalities is identified as a priority issue, the propor- issues when measuring them. Additionally, various tion of pedestrians walking on safe sidewalks is an examples for sets of internationally comparable SPIs important safety performance indicator. It is useful to are introduced. formulate a target for this proportion. First, however, one must define how this can be measured and know While monitoring progress in road safety in the present level to formulate a target (WHO 2013). terms of crashes and casualties is important, because these two indicators occur as the “worst The World Health Organization (WHO) has led a case” of unsafe operational conditions in the road process of developing a set of voluntary global traffic system, they represent only a small portion of performance targets, involving member states and meaningful measurements. SPIs are indicators with key stakeholders (WHO 2017). This process culmi- strong causal links to road safety that better reflect nated in a set of 12 voluntary global road safety per- road safety management and the state-of-the-art in a formance targets, listed in table 2.2. Each target rep- given country (Bliss and Breen 2009; 2013). resents a specific goal to be achieved at a global level, based on the combined efforts of individual countries wishing to contribute toward global objectives. Table 2.2. WHO Global Road Safety Performance Targets Global road safety targets 1 National action plan By 2020, all countries establish a comprehensive multisectoral national road safety action plan with time-bound targets. 2 Global alignment By 2030, all countries accede to one or more of the core road safety-related United Nations legal instruments. 3 New roads By 2030, all new roads achieve technical standards for all road users that take into account road safety or meet a three star rating or better. 4 Existing roads By 2030, more than 75% of travel on existing roads is on roads that meet technical stan- dards for all road users that take into account road safety. 5 Vehicle standards By 2030, 100% of new (defined as produced, sold or imported) and used vehicles meet high quality safety standards, such as the recommended priority United Nations Regulations, Global Technical Regulations, or equivalent recognized national performance requirements. 6 Speeding By 2030, halve the proportion of vehicles traveling over the posted speed limit and achieve a reduction in speed-related injuries and fatalities. 32 MOBILITY AND TRANSPORT CONNECTIVITY SERIES Global road safety targets 7 Motorcycle helmets By 2030, increase the proportion of motorcycle riders correctly using standard helmets to close to 100%. 8 Vehicle occupant By 2030, increase the proportion of motor vehicle occupants using safety belts or standard protection child restraint systems to close to 100%. 9 Driving under the By 2030, halve the number of road traffic injuries and fatalities related to drivers using influence alcohol, and/or achieve a reduction in those related to other psychoactive substances. 10 Distraction by By 2030, all countries have national laws to restrict or prohibit the use of mobile phones mobile phone while driving. 11 Professional drivers By 2030, all countries to enact regulation for driving time and rest periods for professional drivers, and/or accede to international/regional regulation in this area. 12 Emergency services By 2030, all countries establish and achieve national targets in order to minimize the time interval between a road traffic crash and the provision of first professional emergency care. Source: WHO 2017. Each target has one to three relevant indicators, A multitude of infrastructural factors contrib- as detailed in Van den Berghe, Fleiter, and Cliff (2020). ute in several ways to causing, avoiding, or mit- Most of the WHO targets concern SPIs 3, 4, 5, 6, 7, 8, igating crashes. One of the many potential ways and 12. However, other targets on crash and casualty to structure safety principles for road infrastructure outcomes (6, 9), and targets on interventions (1, 2, 10, are those defined in the Dutch “Sustainable Safety” and 11), show how important road safety data are at philosophy: functionality of roads (that is, to generate these different levels. a hierarchically structured road network); assuring homogeneity of masses as well as speed and direc- Performance indicators can relate to any part of tion along a stretch of road; forgivingness of the the road traffic system and are usually classified road environment; and predictability of road course as: directed infrastructure, vehicle/s used, road user and road user behavior by recognizable road design behavior, and post-crash care. (Wegman, Aarts, and Bax 2008). INFRASTRUCTURE As an illustration of useful infrastructure safety The design and operational conditions of the road performance data, the star rating developed by the network play a significant role in the causation International Road Assessment Programme (iRAP) of crashes. Recent safety philosophies—such as the provides useful guidance to assess the safety of indi- Safe System Approach—have thus extended the for- vidual road sections (see figure 2.7). merly user-centered approach to explaining crashes, known as “user blame” and advocate a shared responsibility in road safety; those who design, build, and maintain infrastructure shall share an (ethical) responsibility to avoid fatalities and serious injuries (see for example, ITF 2016). 33 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS Figure 2.7. The International Road Assessment Programme’s Star Rating Source: International Road Assessment Programme website: https://www.irap.org. 34 MOBILITY AND TRANSPORT CONNECTIVITY SERIES Assessments of the highest risk parts of the net- Asset databases can be an important source of work have proven to be very useful in low- and information as well. The basic purpose of pavement middle-income countries (LMICs). These assess- management systems is to develop priority programs ments identify the level of risk for different road user and schedules so that construction, rehabilitation, groups to help identify and prioritize interventions and maintenance work is conducted cost effectively. that deliver cost-effective outcomes. The outcomes Local design data, such as traffic flows, material prop- have led to better understanding and decision mak- erties, and unit costs, allow technical as well as eco- ing regarding effective interventions. The data col- nomic evaluations. Network properties or attributes, lected give some very powerful information for addi- such as shoulder design, pavement texture, and tional analysis beyond the road protection score (RPS) roughness have been shown to affect road safety, or star rating. More than 70 specific variables can be and the combination of the road safety management used individually or in combination to form measur- system with the pavement management system can able performance indicators. To return to the exam- help optimize design, construction, and maintenance ple of pedestrian safety, one can derive the percent- for road safety purposes (see, for example, Tighe, age of the network above 30 kilometers per hour that Cowe Falls, and Morrall 2001). have sidewalks where pedestrians are present. Box 2.3 provides an example of an asset database in the Philippines, maintained by the Department of Public Works and Highways (DPWH). 35 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS Box 2.3 Department of Public Works and Highways, Philippines In the Philippines, the Department of Public Works and Highways (DPWH) maintains an extensive Roads and Bridges Inventory database (figure B2.3.1), which contains georeferenced data on various road infra- structure data elements. Examples of these data elements include road classifications, road conditions, surface type, carriageway width, and annual average daily traffic for the entire national road network in the country. In addition, the DPWH has extensive International Road Assessment Programme (iRAP) data throughout the country, used to guide road safety interventions. Figure B2.3.1 Screenshot of the DPWH Road and Bridge Inventory Database Source: DPWH Roads and Bridges Inventory: https://www.dpwh.gov.ph/dpwh/2020%20DPWH%20Road%20and%20Bridge%20Inventory/index.htm. 36 MOBILITY AND TRANSPORT CONNECTIVITY SERIES VEHICLES ROAD USER BEHAVIOR Vehicle technology can help both reduce the like- In principle, performance indicators should be tai- lihood of crashes and mitigate their severity. Two lored to the problems of the country under review. dimensions of vehicle safety can be distinguished: Nevertheless, some behaviors have proven prob- lematic for road safety in many countries. Some • Passive (also referred to as secondary) safety typical problems, along with the data collected to features such as seatbelts, airbags, and general identify the extent of the problem and monitor crashworthiness of vehicles; and progress, will be discussed in this section. As many of these SPIs require roadside surveys, appendix C • Active (primary) safety features, such as ABS (anti- indicates issues that require attention when conduct- lock braking systems) or ESC (electronic stability ing roadside surveys. In principle, data on observed control). behavior preferred over questionnaires measuring behavior as reported by the respondents. The latter A substantial part of reductions in fatalities over are subject to awareness, memory, and social desir- the past decades can be attributed to improve- ability issues. Nevertheless, questionnaires can be a ments in increased active and passive vehicle cost-efficient alternative to collecting information on safety (European Commission 2018). For example, several types of behavior at once. Méndez et al. (2010) show in the case of Spain that drivers of cars registered before 1985 have a signifi- A set of comparable indicators is provided by the cantly higher probability of being killed or seriously E-Survey of Road Users’ Attitudes (ESRA) project. injured than drivers of cars registered between 2000 ESRA (https://www.esranet.eu/en/) is a joint interna- and 2005. tional initiative of research centers and road safety institutes in 38 countries across five continents. The The Global New Car Assessment Program (NCAP) purpose of this network is to collect comparable data has evolved as the most widely used measure for on the road safety situation and culture as indicated the crashworthiness of passenger cars. Originally by road users’ self-reported habits and behaviors, tailored to assess prevalence and function of pas- attitudes, beliefs, perceived norms and values. The sive safety features, credits in the rating (1 to 5 stars) data collected by means of online surveys yield a since 2009 are also given for active safety features, large set of road safety indicators. Although behav- that is, devices intended to reduce the probability of ior observations are always preferred above self-re- a crash. The NCAP safety rating today is composed ported behavior measurements, the ESRA question- of tests in four categories: adult occupant protection, naire is a relatively cheap way to collect information child occupant protection, pedestrian protection, across countries on all the behaviors mentioned and safety assist. The latter is dedicated to advanced above at the same time and in a comparable way. driver assistance systems (ADAS) and includes speed assistance, automatic emergency braking, and lane support, among others (European Commission 2018). Monitoring the NCAP distribution of a country’s fleet helps evaluate efforts to improve the fleet’s safety, such as new car incentive systems (see box 2.4 for details of the NCAP in action in Southeast Asia). 37 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS Box 2.4 Malaysian Institute for Road Safety Research and the Association of Southeast Asian Nations New Car Assessment Program The Malaysian Institute for Road Safety Research (MIROS), as the road safety center for the Association of Southeast Asian Nations (ASEAN), has been in the forefront of road safety research and policy in South- east Asia since 2012. One of its main activities is partnering with the ASEAN New Car Assessment Program (NCAP), which implements vehicle safety standards in the region. Vehicles that undergo the collision tests and meet the standards set by the ASEAN NCAP also meet the United Nations vehicle safety standards. As hosts of the collision testing laboratory (figure B2.4.1), MIROS collects good quality vehicle safety data in the region, which are used for road safety policies and programs. Currently, MIROS is improving their test- ing capabilities to also include safety technologies for motorcycle safety. Figure B2.4.1 ASEAN NCAP Collision Test Conducted in the MIROS Testing Laboratory Source: MIROS (https://miros.gov.my). See the ASEAN NCAP website for information about the 2013 collision test for the Proton Prevé shown in the image: https://aseancap.org/v2/?p=3448. 38 MOBILITY AND TRANSPORT CONNECTIVITY SERIES SPEEDING is available on sample composition. Traditionally, Excessive and inappropriate speed is the most speed measurements have been based on roadside critical factor contributing to road injuries in measurements with radars, cameras, or tubes. These many countries. The higher the speed, the greater methods are comparable over time, if the devices stay the stopping distance required, a more diminished at the same location, which is important for monitor- field of vision, and hence the increased risk of a crash. ing progress. However, the representativeness of the As more kinetic energy must be absorbed during a measurement location can be an issue and in gen- high-speed impact, the risk of injury increases should eral the density of the measurement points will not a crash occur. Heterogeneity of driven speeds is also be sufficient to match local speeds to crashes. SPIs a risk factor as it increases the likelihood of con- typically reported include the mean speed, V85 (the flicts. Traditionally, the identification of pre-crash speed not exceeded by 85 percent of drivers) and the speed requires a crash reconstruction often beyond speed variance (GRSP 2008). the routine investigations conducted by the police. Therefore, coding of speed as a possible causation HELMET USE factor is often unreliable. Event data recorders that The growing fleet of motorized two-wheelers in register the speed and acceleration in the seconds many LMICs makes the requirement of helmet up to the impact are increasingly common. If the use an urgent priority. Head injuries among users police have the materials and expertise to read these of two-wheelers is a primary concern. Motorcycle and records (requiring a capacity only starting to develop bicycle helmets effectively prevent head injuries and in most countries), these data allow reliable conclu- reduce the severity of injuries sustained by riders and sions about the role of speed in crash causation. SPIs, passengers of two-wheelers. However, data on head with respect to speeding, can be based on mean injuries resulting from motorcycle crashes might be speeds available from various sources—such as big difficult to interpret, for instance, because the dis- data providers including Uber, Google, and others— tances traveled by motorcycle are unknown and rap- or in more detail as sold by providers of navigation idly changing. An observational study can, therefore, services. These data are also called floating car data help estimate the proportion of motorcyclists who and from these, speed indicators can be meaning- wear helmets, and can provide a better way of test- fully calculated for road sections—at least for rural ing the success of a program to increase helmet use roads—which allows direct linking to crash counts. (WHO 2006). Collecting this information in roadside However, stability over time is an issue, because the surveys is relatively straightforward (see appendix C). samples can vary strongly and usually no information Alternatively, this can be covered by a survey. 39 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS SEATBELTS AND CHILD RESTRAINT SEATS Collecting this information in roadside surveys is rela- One of the most effective measures to protect tively straightforward (see appendix C). Alternatively, occupants from injury in a crash is the proper this can be covered by a survey. fitting and use of seatbelts and child restraints. Both are proven to save lives and reduce injury DRIVING UNDER THE INFLUENCE severity, and all vehicle occupants should be appro- The consumption of alcohol, even in relatively priately restrained when traveling in a motor vehicle. small amounts, increases the risk of being Worldwide, however, not all vehicles are fitted with involved in a crash for all types of road users. In seatbelts, and not all occupants use them when they many countries, research indicates that considerable are available. In countries where car use is rising proportions of drivers, motorcyclists, and pedestri- most rapidly, the use of seatbelts and child restraints ans have alcohol in their blood in sufficient concen- is low. In addition, outcome data on seatbelt use is trations to impair road user awareness. However, often unreliable because once the driver or passen- good data on drink driving is difficult to attain. Breath ger has left the car, investigating officers do not know or blood tests of road users involved in a crash whether the seatbelt was being used. For this rea- are not always conducted—especially if they have son, it is often more reliably coded for fatalities than life-threatening injuries. This is particularly problem- for injured casualties. An indirect way to monitor the atic because drink driving increases the risk of fatal effect of seatbelts is the analysis of ejections from the crashes much more than the risk of lighter crashes. vehicle or of injuries due to occupants striking the Therefore, indicators mainly based on drink driving in windscreen, both of which are less likely when a seat- light crashes leave out an important part of the prob- belt was worn. Performance indicators for the use of lem. In terms of SPIs, the number of alcohol-related seatbelts and child restraint seats include the propor- offenses detected by police gives a reasonably com- tion of vehicles fitted with seatbelts (in all seating plete picture over time—unless enforcement efforts positions), the proportion of car occupants using change. The best practice, though difficult to carry a seatbelt, and the proportion of children (correctly out, is random testing of drivers in a road-side survey fastened) in child restraints (FIA Foundation 2009). (GRSP 2007). 40 MOBILITY AND TRANSPORT CONNECTIVITY SERIES DISTRACTION PROFESSIONAL DRIVERS The rise in mobile communication devices and the Driving requires continuous attention to the information and entertainment they offer to driv- road, other road users, and your own vehicle. ers has made distraction a likely determinant for Long-term driving without a break can reduce atten- rises in road crash casualties in many countries tion and prolong reaction time, increasing the like- (ITF 2020). An American naturalistic driving study esti- lihood of crashes. As an example, as compared to a mates in more than half of all driving situations, the driver who has driven between 8 and 9 hours, the driver was distracted and on average the crash-risk of crash risk in the tenth hour of driving is increased by a distracted driver is doubled as compared to a driver 70 percent and in the eleventh hour by 300 percent. who is fully attentive. For activities such as texting, One break reduces the crash risk by 18 percent and this risk rises to 10 times as high as for a nondiverted two breaks by 37 percent (Hoye 2016). Fatigue due to driver (Dingus et al. 2016). With respect to distraction, extended periods of driving can be a particular prob- crash and performance data are, however, difficult to lem for professional drivers who often have to drive obtain. Event data recorders will help identify drivers long hours without breaks to maximize their often who never tried to prevent the crash by steering or meager earnings. This is the case in countries where braking, making either distraction or fatigue a likely public transport is privatized, with no laws to regulate causation factor; however, until the readout of these the driving times. Questioning professional drivers data has become common practice, a broad identifi- about their driving times and any crashes they were cation of crashes caused by distraction is impossible. involved in will help to shed light on this problem. If Roadside observations can identify drivers holding legislation on driving times and rest periods for pro- a cellphone, and can even be automated with mod- fessional drivers exist, equipping the vehicles used ern video analytical software. Note, however, that by professional with a tachograph is an important according to Dingus and colleagues (2016), 78 per- first step; the percentage of vehicles with a tacho- cent of distracting activities do not involve holding graph would serve as the resulting indicator. The a cell phone. The evaluation of distracted driving in percentage of drivers whose tachographs have been questionnaires is also problematic, as this behav- checked, and the percentage of drivers who comply ior is strongly susceptible to awareness and mem- with requirements for driving time and rest periods ory effects. Questions about this behavior are most would become the indicators for the next steps. reliably answered when they relate to a trip that the respondent has just finished or interrupted, for In Europe, the Baseline Project (box 2.5) is helping to instance, at a rest stop. pinpoint road safety issues through a set of key per- formance indicators (KPIs). 41 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS Box 2.5 European Road Safety: The Baseline Project The European Commission of the European Union (EU) has elaborated a list of key performance indicators (KPIs) for road safety to gain a clearer understanding of the different issues influencing overall safety per- formance. This minimal set of performance indicators has been selected in close cooperation with experts and authorities from the EU member states as a compromise between optimal information and practical feasibility (figure B2.5.1). The Baseline Project (https://baseline.vias.be) aims to collect a set of harmonized indicators based on a common methodology. This methodology (preliminary versions were published in May 2021) will update older European manuals for the collection of SPIs (Hakkert, Gitelman, and Vis 2007). Figure B2.5.1 The Baseline Project: List of KPIs for Road Safety, with Definitions Source: The Baseline Project website: https://baseline.vias.be; European Commission 2019; Hakkert, Gitelman, and Vis 2007. POST-CRASH CARE first strategy aims at the quickest possible transfer The so-called “golden hour” in trauma manage- to the hospital, the second aims more at treating ment refers to the hypothesis that a patient has and especially stabilizing the patient before trans- the best chances of survival when transported port, accepting (when necessary) a longer time spent to a hospital within an hour after the crash. on-scene (Johannsen et al. 2017). The most basic Consequently, paramedics are taught to work as indicator is the percentage of injured casualties that fast as possible on-scene so the “golden hour” is receive professional health care. If the scoop-and-run not exceeded, despite the different philosophies in strategy is employed, the time elapsed between the trauma care. One can roughly differentiate between crash and arrival in hospital becomes the most rele- two principles in management: “scoop and run,” a vant indicator. For the second strategy, stay and play, strategy preferred in the United States, Poland, and the time to first contact with professional emergency the United Kingdom, and “stay and play,” practiced care is the more relevant indicator. in Germany and France, among other countries. The 42 MOBILITY AND TRANSPORT CONNECTIVITY SERIES 2.4. Data on Road Safety Interventions Road safety interventions range across road and only detailed implementation data will permit the vehicle engineering, enforcement, education, evaluation of the effectiveness of a measure in terms training and promotion, as well as medical ser- of reducing casualties or risky behavior. Nonethe- vices for crash casualties. Follow-up indicators for less, sometimes when a measure has been adopted, interventions are an important part of monitoring the no effect can be seen. In that case, it is important to implementation of countermeasures. know whether the measure was duly implemented and did not work, or whether barriers prevented Road safety management should follow the steps implementation. of a plan-do-check-act (PDCA) process (ITF 2016). To follow up the measures taken to improve road Bliss and Breen (2009), therefore, suggest monitoring safety, it is not only important to monitor progress safety interventions with regard to their output quan- in terms of outcomes (crash data) and safety per- tities. See figure 2.8 for areas of interventions and formance (SPI data), but also to follow up the actual examples for indicators. implementation of the agreed measures. Moreover, Figure 2.8. Intervention Indicators, Main Fields and Examples • Road sections with improved iRAP star rating Safety engineering • Number of intersections improved • Number of speed camera operational • Length of road with section control for speeding • Number of tickets delivered Enforcement • Number of drivers checked • Hours spent on check • Number of downloads for educational material Education • Number of children taught a course Promotional activities • Number of clicks on promotional video • Minutes of air time for a spot Driver training • Driving lessons taken by students • Exams attempted/exams passed Vehicle testing • Vehicles checked • Vehicles admitted after improvements Emergency medical • Crash scenes attended services • Average time to arrive at scene Source: Adapted from Bliss and Breen 2013. 43 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS 2.5. Reports and Data from Regional Road Safety Observatories Regional road safety observatories in Latin Amer- the reviewer guidance on road safety priorities in the ica, Asia-Pacific, and Africa, and other regions can region and, in turn, identify which types of data are also provide regional road safety data that will offer most crucial and needed to improve road safety in a further context in a national data review. Through country. For example, motorcycle crashes are one of the reports and outputs of the observatories, the the most pressing problems in the Asia-Pacific region performance of a particular country in terms of data and will require more robust data in terms of motor- collection can be compared with other similar coun- cycle safety. tries in the region. Observatories can also provide 44 MOBILITY AND TRANSPORT CONNECTIVITY SERIES References AAAM (American Association for Automotive Medicine). 2016. The Abbreviated Injury Scale © 2005. Update 2008, edited by T. Gennarelli, and E. Woodzin. Chicago, IL: AAAM. https://www.aaam.org/abbreviated-injury-scale-ais/. Antoniou, Constantinos, and George Yannis. 2013. “Assessment of Exposure Proxies for Macroscopic Road Safety Predic- tion.” Transportation Research Record: Journal of the Transportation Research Board 2386 (1): 81–94. https://doi.org/10.3141/2386-10. Austroads. 1997. A Minimum Common Dataset for the Reporting of Crashes on Australian Roads. Sydney, N.S.W.: Austroads. https://nla.gov.au/nla.cat-vn357058. Azzouzi, Mustapha. 2019. “Survey Analysis Report—First Results: Evaluating Country Level Adherence to Crash-Related Vari- ables.” Presentation at the SSATP Annual General Meeting, Victoria Falls, Zimbabwe, November 26. https://www.ssatp.org/publication/survey-analysis-report-first-results-evaluating-country-level-adherence-crash-related. Bliss, Tony, and Jeanne M. Breen. 2009. Country Guidelines for the Conduct of Road Safety Management Capacity Reviews and the Specification of Lead Agency Reforms, Investment Strategies and Safe System Projects. Washington, D.C.: World Bank Group. http://documents.worldbank.org/curated/en/712181469672173381/GRSF-Country-Implementation-Guidelines. Bliss, Tony, and Jeanne Breen. 2013. “Road Safety Management Capacity Reviews and Safe System Projects Guide- lines.” Updated edition. Global Road Safety Facility, Washington, D.C. http://documents.worldbank.org/curated/ en/400301468337261166 Burlacu, Alina F. 2019. “DRIVER (Data for Road Incident Visualization, Evaluation, and Reporting): The World Bank’s Tool for Crash Data Visualization, Evaluation and Reporting.” Presentation at the ITF/ESCAP Meeting, “Supporting Traffic Safety Information Systems of Countries in Southeast Asia,” Bangkok, Thailand, June 14. https://www.itf-oecd.org/node/23935. CARE Team. 2018. CARE Database: Common Accident Data Set [CADaS] Reference Guide. Version 3.7. Directorate-General for Mobility and Transport, European Commission. https://ec.europa.eu/transport/road_safety/sites/roadsafety/files/ cadas_glossary_v_3_7.pdf. Dingus, Thomas A., Feng Guo, Suzie Lee, Jonathan F. Antin, Miguel Perez, Mindy Buchanan-King, and Jonathan Hankey. 2016. “Driver Crash Risk Factors and Prevalence Evaluation Using Naturalistic Driving Data.” Proc. Natl. Acad. Sci. U.S.A. 113 (10): 2636–41. https://doi.org/10.1073/pnas.1513271113. Duchamp, Gilles. 2017. Bases de données pour la sécurité routière : Principes, enjeux, recommandations. Presentation at the Africa Transport Policy Program (SSATP) & International Traffic Safety Data and Analysis Group (IRTAD) Workshop on Road Safety Data, Marrakech, Morocco, February 23–24. https://www.ssatp.org/sites/ssatp/files/annual_meetings/2017/ Presentations/Road_Safety_Workshop/Session%204-Bases%20de%20donn%c3%a9es%20pour%20la%20s%c3%a9cu- rit%c3%a9%20routi%c3%a8res.pdf. European Commission. 2018. Monitoring Road Safety in the EU: Towards a Comprehensive Set of Safety Performance Indicators 2018. Directorate General for Transport, European Commission. https://ec.europa.eu/transport/road_safety/sites/ default/files/pdf/ersosynthesis2018-performanceindicators.pdf. 45 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS European Commission. 2019. “Commission Staff Working Document EU Road Safety Policy Framework 2021–2030: Next Steps Towards ‘Vision Zero.’” Working Paper, European Commission, Brussels, Belgium. https://ec.europa.eu/transport/road_safety/sites/roadsafety/files/1_en_document_travail_service_part1_v2.pdf. Eurostat, ITF (International Transport Forum), and UNECE (United Nations Economic Commission for Europe). 2019. Glossary for Transport Statistics. 5th ed. Luxembourg: Publishing Office of the European Union. https://doi.org/10.2785/675927. FIA Foundation (FIA Foundation for the Automobile and Society). 2009. Seat-Belts and Child Restraints: A Road Safety Manual for Decision-Makers and Practitioners. London: FIA Foundation. https://www.who.int/roadsafety/projects/manuals/seatbelt/en/. GRSP (Global Road Safety Partnership). 2007. Drinking and Driving: A Road Safety Manual for Decision-Makers and Practi- tioners. Geneva, Switzerland: Global Road Safety Partnership. https://www.grsproadsafety.org/wp-content/uploads/ DrinkingDriving_English.pdf. GRSP (Global Road Safety Partnership). 2008. Speed Management: A Road Safety Manual for Decision-Makers and Practitioners. Geneva, Switzerland: Global Road Safety Partnership. https://www.who.int/publications/i/item/ speed-management-a-road-safety-manual-for-decision-makers-and-practitioners. Hakkert, A. S., V. Gitelman, and M. A. Vis, eds. 2007. “Road Safety Performance Indicators: Theory. Deliverable D3.6 of the EU FP6 Project Safetynet.” Report, Loughborough University, United Kingdom. https://hdl.handle.net/2134/4952. Hoye, Alena. 2016. Regulering av kjøre- og hviletid. Trafikksikkerhetshåndboken 6.10. Institute of Transport Economics, Norway. https://www.tshandbok.no/del-2/6-krav-til-foerere-foereropplaering-og-yrkeskjoering/doc719/. ITF (International Transport Forum). 2015. Why Does Road Safety Improve When Economic Times Are Hard? Paris, France: Organisation for Economic Co-operation and Development/International Transport Forum. https://www.itf-oecd.org/ why-does-road-safety-improve-when-economic-times-are-hard. ITF (International Transport Forum). 2016. Zero Road Deaths and Serious Injuries: Leading a Paradigm Shift to a Safe System. Paris, France: OECD Publishing. http://dx.doi.org/10.1787/9789282108055-en. ITF (International Transport Forum). 2020. Road Safety Annual Report 2020. Paris, France: Organisation for Economic Co-op- eration and Development/International Transport Forum. https://www.itf-oecd.org/road-safety-annual-report-2020. Johannsen, H., N. O’Connell, J. Ferrando, K. Pérez. 2017. “Prehospital Care, European Road Safety Decision Support System,” developed by the Horizon 2020 project SafetyCube, European Commission, Brussels, Belgium. https://www.roadsafe- ty-dss.eu; https://www.roadsafety-dss.eu/assets/data/pdf/synopses/Prehospital_Care_30012018.pdf. Kopits, E. and M. Cropper.2005. “Traffic Fatalities and Economic Growth.” Accident Analysis and Prevention 37 (1): 169–178. http://dx.doi.org/10.1016/j.aap.2004.04.006. Martensen, Heike. 2018. “Data Review Mission in Serbia.” Internal working paper prepared for the International Traffic Safety Data and Analysis Group (IRTAD), ITF/OECD, Paris, France. Méndez, Á. G., F. A. Izquierdo, and B. A. Ramírez. 2010. “Evolution of the Crashworthiness and Aggressivity of the Spanish Car Fleet.” Accident Analysis and Prevention 42 (6): 1621–31. https://doi.org/10.1016/j.aap.2010.03.020. 46 MOBILITY AND TRANSPORT CONNECTIVITY SERIES Papadimitriou E., and G. Yannis. 2018. “Needs and Use of Road Safety Data within the UN SafeFITS Model.” Presented at the Albania Road Safety Performance Review Capacity Building Workshop, Durres, Albania, February 6–7. https://www.nrso.ntua.gr/geyannis/wp-content/uploads/geyannis-cp281.pdf. PIARC (World Road Association). 2019. “Chapter 5.2: Identifying Data Requirements.” In Road Safety Man- agement: Road Safety Manual (Part II). La Défense, CEDEX, France: PIARC. https://roadsafety.piarc.org/en/ road-safety-management-safety-data/identify-data-requirements. Shinar, D., P. Valero-Mora, M. van Strijp-Houtenbos, N. Haworth, A. Schramm, Guido De Bruyne, V. Cavallo, J. Chliaoutakis, J. Dias, O .E. Ferraro, A. Fyhri, A. Hursa Sajatovic, K. Kuklane, R. Ledesma, O. Mascarell, A. Morandi, M. Muser, D. Otte, M. Papadakaki, J. Sanmartín, D. Dulf, M. Saplioglu, G. Tzamalouka. 2018. “Under-Reporting Bicycle Accidents to Police in the COST TU1101 International Survey: Cross-Country Comparisons and Associated Factors.” Accident Prevention and Analy- sis 110 (January): 177-186. https://doi.org/10.1016/j.aap.2017.09.018. Tighe, Susan, Lynne Cowe Falls, and John Morrall. 2001. “Integrating Safety with Asset Management Systems.” Paper No. 63, Submission for the 5th International Conference on Managing Pavements, Seattle Washington, August 11–14. https://www.researchgate.net/publication/228418327. Van den Berghe, W., J. J. Fleiter, and D. Cliff. 2020. Towards the 12 Voluntary Global Targets for Road Safety: Guidance for Coun- tries on Activities and Measures to Achieve the Voluntary Global Road Safety Performance Targets. Brussels, Belgium: Vias institute and Genève: Global Road Safety Partnership. https://www.vias.be/publications/Towards%20the%2012%20Volun- tary%20Global%20Tar-gets%20for%20Road%20Safety/Towards_the_12_Voluntary_Global_Targets_for_Road_Safety.pdf. Wegman, Fred, Letty. Aarts, and Charlotte. Bax. 2008. “Advancing Sustainable Safety. National Road Safety Outlook for The Netherlands for 2005–2020.” Safety Science 46 (2) 323–343. https://doi.org/10.1016/j.ssci.2007.06.013. Wegman, Fred. 2016. “Chapter 4: Road Safety Data Collection, Analysis, Indicators and Targets.” In Halving the Number of Road Deaths in Korea: Lessons from other Countries, 83–102. Paris, France: OECD/ITF. https://www.itf-oecd.org/sites/ default/files/docs/halving-road-deaths-korea.pdf. WHO (World Health Organization). 2006. “Helmets: A Road Safety Manual for Decision-Mak- ers and Practitioners.” WHO, Geneva, Switzerland. https://www.who.int/publications/i/item/ helmets-a-road-safety-manual-for-decision-makers-and-practitioners. WHO. 2010. “Data Systems. A Road Safety Manual for Decision-Makers and Practitioners.” WHO, Geneva, Switzerland. https://www.who.int/publications/i/item/data-systems. WHO. 2013. Pedestrian Safety: A Road Safety Manual for Decision-Makers and Practitioners.” WHO, Geneva, Switzerland. https://www.who.int/publications/i/item/pedestrian-safety-a-road-safety-manual-for-decision-makers-and-practitioners. WHO. 2017. “Global Road Safety Performance Targets.” WHO, Geneva, Switzerland? https://www.who.int/violence_injury_ prevention/road_traffic/12GlobalRoadSafetyTargets.pdf. Yannis, George, E. Papadimitriou, P. Lejeune, V. Treny, S. Hemdorff, R. Bergel, M. Haddak, P. Holló, J. Cardoso, F. Bijleveld, S. Houwing, and T. Bjørnskau. 2005. “State of the Art of Risk and Exposure Data.” Deliverable 2.1 of the EC FP6 project SafetyNet, European Commission, Brussels, Belgium. https://www.swov.nl/en/publication/ state-art-report-risk-and-exposure-data. 3. Preparing for the Road Safety Data Review 48 MOBILITY AND TRANSPORT CONNECTIVITY SERIES During a road safety data review, the practice of When assessing the situation in the target coun- different actors should be assessed within the coun- try, the reviewer should identify gaps between try and the adopted approaches relative to inter- the best international practices, as summarized national standards. To prepare for a road safety in the previous chapter, and which steps have been data review, a realistic scope for the topics to be completed, as reported by the country’s institutions. addressed should be agreed upon. The host team More importantly, the reviewer should confirm should collect information and arrange meetings whether the official procedure is also implemented in ahead of the visit. The review team should analyze practice. The situation assessment process is repre- all information made available and look for addi- sented in figure 3.1 as a Venn diagram. tional information, such as earlier reviews or inter- national publications. Figure 3.1. Situation Assessment: Comparison between the Official Procedure (Declarations), the Actual Practice (Facts), and the Internationally Recommended Practices (International References) International references: what should be done 4 2 3 1 7 5 6 Declarations: Facts: what is said to be done what is really done Source: Original figure produced for this publication. 49 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS The reviewer should situate each aspect of the • Area 6: The procedure contains certain actions, road safety data system in one of the six areas and these have been put into practice. Before symbolized in this Venn diagram and adapt their commenting on these, the reviewer should check recommendations accordingly. carefully whether each action has a purpose spe- cific to the target country. • Area 1: This is the ideal situation; the reviewer needs to simply provide a certificate of conformity. • Area 7: The official procedure contains actions for which the purpose is unclear, but these have not • Area 2: In this area, the stakeholders understand been put into practice. The procedure could be what must be done, and state it has been com- adjusted to increase its credibility. pleted; however, the reviewer sees that actual practice differs from the official declaration. Finally, the reviewer should strive for full imple- Nonconformities in the procedure do not result mentation of the necessary procedure that leads from lack of knowledge, but from implementa- to data conforming to international standards, or tion issues, including those related to materials, the green circle. The review therefore focuses on the organization, training, and budget. The recom- implementation of the necessary actions to achieve mendations will therefore focus on practical that goal, notably areas 1 through 4 in the diagram. implementation. A data review can serve several purposes and • Area 3: Those in charge do not fully understand should reflect the level of development in road whether their procedural practice is sound. The safety data within a country. Objectives for the reviewer should investigate why this discrepancy review can include: (a) working together on road exists, present what is done as good practice, safety data to build initial trust, (b) building capacity, and decide whether the practice should be kept, (c) empowering a team in the host country, and (d) revised, or increased. facilitating progress with data as well as other road • Area 4: This is the area for real progress: While safety initiatives. Depending on the context, the road not yet completed (neither declared nor in prac- safety data review can serve as the starting point of tice), the procedure has been deemed essential. this process or take place at a more advanced stage. The reviewer will first have to insist on the need In the first case, the visit mainly serves for consul- to implement the intended practice. Focus on the tation and getting to know the actors and helps to necessary training to make those involved aware establish a mentoring system where a team in the of the importance of this action. country is set up and mentored by one or more experts over a longer period. Further work can • Area 5: In practice, certain actions have been continue with online meetings and could be linked undertaken; however, these actions are not part with the capacity-building programs planned in the of the official procedure and their purpose is not observatories. clear. Stakeholders should consider committing resources to the procedure that could be invested more usefully. 50 MOBILITY AND TRANSPORT CONNECTIVITY SERIES If data collection is already more advanced, the visit Because a broad range of different data types can serve as a review of existing practice. The review are all relevant for road safety management, it report will describe the strengths and weaknesses is important to define the scope of the review of the data-collection system and indicate where beforehand. Detailed discussions could be required improvements can be made. to address all the issues mentioned here and collect all relevant information. Language barriers, for exam- Challenges triggered by COVID-19 made online ple, can make discussions quite lengthy and a federal meetings much more common and demonstrated structure might require assessing the practice in dif- many discussions can be successfully led online. ferent states. Revising the line of questioning accord- Online meetings can therefore form an integral part ing to the first discussions might prove necessary, of the preparation and the follow-up. However, the which could reduce or extend the scope of the review key objectives mentioned above—trust building and based on progress with the interviews. However, in investigating the actual practice in data-collection— principle, a clear scope should be defined before- can require in-person meetings. Sensitive issues hand. Depending on the scope of the review, different should preferably not be discussed online, especially stakeholders will need to be contacted. For example, if several persons are involved. Technical discussions if indicators for road quality have to be investigated, play an important part of building personal relations, the reviewer must meet engineers in the transport but not the only one. Spending time together also departments responsible for construction and main- helps review participants get to know each other tenance of the road network, while also evaluating the on a more personal level, which facilitates future collection of police enforcement data, which would cooperation. require meetings with police and justice departments. It takes about a week to investigate the collec- 3.1. Scope of the Review tion, storage, and use of crash and casualty data. Safety performance indicators (SPIs) and mobility data comprise a complex field because they typically If time is limited, defining which data should be include many different data types and each could reviewed becomes especially important. For an ini- require similar efforts to investigate them as do the tial visit of approximately one week, the focus would crash and casualty data. Consequently, a thorough be on crash data; however, the review should also investigation of the whole data collection process— assess the availability of other data. A follow-up visit as foreseen for crash and casualty data—might take could be required. Depending on several factors, a another week. Thus, the review should focus on crash large part of the week should be reserved for the full and casualty data first. For SPI and mobility data, coverage of crash and casualty data across the whole the emphasis could be on discussing which indica- chain, from observation continuing through to the tors and mobility data are available and how each is making information available to all stakeholders. The linked to crash data and used in the analyses. Based general availability and use of other data types can on the problems identified in the crash data and on be addressed in a much shorter time. the discussions with various stakeholders, the review team will gain some insights into which type of indi- cators would be useful and feasible. 51 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS 3.2. Preparations by the Host Team The hosting team should prepare information on For the review team to adequately prepare, the deliv- different aspects of road safety data collection: ery date should be agreed upon in advance of the (a) an overview of the indicators available, (b) mission. Documents should preferably be provided in the actors involved, (c) a description of the data the language of the review, but should also be sent in collection procedure for the indicators defined the country’s language, if possible. in the scope, and (d) examples of the use of the data. Before the visit takes place, the hosting team The following section explains in greater detail the should help the review team arrange meetings with documentation required, along with some practical all important stakeholders. considerations on how to organize the visit. The host country should be represented by the OVERVIEW lead agency for road safety, often the national The host country should produce a table of all steering committee, a transport-related ministry relevant road safety data available (see table department, the national road agency, or other 3.1). This table should include crash and casualty appropriate agencies. Ideally, this agency should also data (from police and hospital records) and, if appli- be responsible for road safety analyses and more cable, also list road safety performance indicators importantly, for coordinating road safety activities and mobility data. The list should indicate for how across all sectors and stakeholders. long (and at what frequency) the data are available and give some important splits that can be made, The following documentation should be provided to for example, by road user type, by age group, and by the review team, if available: road type. The institution in charge of collecting and maintaining the data should also be indicated. • Overview of all available road safety data If available, documentation about the data collec- • List of key partners in road safety data collection, tion process should be sent ahead. As an example, analysis, and use the collection of crash data should ideally start with • Crash registration form (police) the officer who records the crash at the scene and end with the crash database. Some questions to be • Registration form (hospital) answered include the following: What are the inter- mediate steps? How are the data collected, aggre- • Definitions and legal basis gated, cleaned, augmented, and checked? How are • Relevant outputs: reports, maps, analyses data stored and organized? How are data shared and analyzed for road safety interventions? • Data tables 52 MOBILITY AND TRANSPORT CONNECTIVITY SERIES Table 3.1. List of Available Road Safety Data History: Important Institution responsible First year (and frequency) variables available for data collection for regular reporting or dates (including age, road user for ad-hoc measurement type, road type) Police/insurance data: Fatalities MAIS 3+a Severely injured Slightly injured Property damage only Hospital data: Emergency room patients Admissions MAIS 3+ Ambulances Exposure data: Travel distances Vehicle fleet Number of valid driver’s licenses Road length Safety Performance Indicators: Helmet usage Seat belt usage Driving under the influence Speeding Distraction Road protection scores: Vehicles’ NCAPb scores: Other: Source: Original table produced for this publication. Note: a. MAIS 3+ = Maximum Abbreviated Injury Scale, where an injury is rated greater than 3 on a scale of 1 (minimal injury) through 6 (maximal injury). Injuries rated 3 and higher are considered critically serious; b. NCAP = New Car Assessment Program. 53 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS KEY PARTNERS IN ROAD SAFETY DATA maintaining the database, staff involved in informa- COLLECTION, ANALYSIS, AND USE tion technology (IT) system architecture and govern- Usually, the relevant organizations that should be ment servers, and possibly staff who check the injury listed include: data and link them to other data sources. • Ministries involved with road safety Data collection can concern police, hospitals, and • Police also the vital registration and mortality statistics. For each of these, the review team should trace who • Hospitals collected the data in the first place, who maintains the system for their storage, and who has access to • Statistics office it. If the data are linked, the responsible party should • Agencies and coroners in charge of vital be indicated. registration Moreover, information should be provided on • Transport/road/public works department how data are being used by the key stakeholders. engineers In this description, the following questions should be • Vehicle registration and driver licensing considered: departments • Who analyzes road safety data, and for whom? • Nongovernmental organizations (NGOs) • How are police using data to direct their activity? • Road safety advocacy organizations • How do engineers use the data? • Academia and research institutes • Are road safety data linked to the design, con- • First responders, such as fire and rescue services struction, and maintenance of the road network? and disaster management teams • Are the crash data linked to asset data? • Insurance companies • Do any other groups use the crash data, such as insurance companies, NGOs, researchers? If subnational authorities play an important role in road safety management, they could also be included on the list. RELEVANT OUTPUTS: REPORTS, MAPS, ANALYSES Ahead of the visit, the review team should try to The main stakeholders and their roles and gain an insight into how available road safety responsibilities in the process of data collec- data are used. Some outputs should be requested tion should be described. The reviewers should as examples of the types of analysis conducted in the follow the path of data collection and talk to every- country under review and which stakeholders are one involved in the whole chain, for example, med- targeted. The host country’s team should indicate ical staff who document their patients’ injuries and whether these outputs are published regularly (for the causes, coders who enter the information into example, annually) or as ad hoc studies. These out- the database, staff responsible for building and puts should help to answer the following questions: 54 MOBILITY AND TRANSPORT CONNECTIVITY SERIES • Which analyses are carried out to inform road • Who qualifies as a casualty? safety policy-making? o How are fatalities, severely injured, and slightly • Which stakeholders are making use of the data? injured defined in the country under review? • Which types of data are used? The legal framework for data registration activ- ities should be described. If in place, legislation • Are there analyses where different types of road should be presented on (a) the investigation of road safety data are combined? crashes, (b) cataloguing different types of vehicles, or (c) national statistics, such as regulating princi- CRASH AND CASUALTY REGISTRATION FORM ples of reliability, independence, transparency, and A crash data form used by police as well as health data protection. Moreover, information should be data forms used by hospitals, along with a glos- provided on whether the obligation to share data is sary for the variables and possible values, should regulated by law or—on the contrary—impaired by be made available as PDFs (portable document for- privacy regulations. mat), if possible. DESCRIPTION OF DATABASE(S) DEFINITIONS FOR CRASH DATA The host country should provide a description International transport statistics are based on of how road safety data are stored. This should the glossary for transport statistics (Eurostat, ITF, contain the type of data, the variables, and the archi- and UNECE 2019). Aligning crash data reporting with tecture of the database. Materials describing the this set of definitions is advised. The reviewer should following aspects of the database should be sent, if check the following: available: • What counts as a road crash? • Variables/glossary o To collect data on road traffic crashes, the • Software, such as Excel or other dedicated review must define which crashes are included software and excluded. For example, the host country • Architecture could indicate for the following cases whether they would be considered as road crash casual- • Checks and controls ties: drivers committing suicide or having heart • Links to other data attacks; single vehicle crashes, crashes without • Access motorized vehicles; and crashes on nonpublic roads. If any national definitions are used by • Output options, such as visualizations, dash- the police, these should be made available. boards, and standard report, among other options 55 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS ACTUAL CRASH DATA VISIT PREPARATION CHECKLIST For the crash data, the review team should check The hosting team should go through the follow- the data for temporal and spatial consistency. ing checklist: Annual data for the period for which data are con- • Appoint a host team associated with the lead sidered valid should be investigated. Possible breaks agency for road safety. The host(s) should help due to changes in methodology should be indicated. identify the stakeholders, accompany the review- The following tables of annual figures should be pro- ers, and, if necessary, translate and interpret. vided, if available: • Organize an interministerial or interagency • Fatalities by age group and gender meeting. • Fatalities by road type • Organize and schedule appointments with rel- • Fatalities by road user type and region evant officers, at different levels, from different agencies, including the following: • Number of severely injured, hospitalized, or Maxi- mum Abbreviated Injury Scale (MAIS) 3+ casualties o Directorate of the ministries involved with road safety; possibly the same as for police: • Percentage of fatalities who died on the spot, as opposed to those who died in hospital – Road safety entity (bureau of this adminis- tration, or mandated institute) that usually • Population by region provides national road safety data analysis – Local entities of the above ministry in charge The review team should also request the number of of local implementation of road safety policy road crash fatalities from vital statistics, recording the occurrence and characteristics of vital events such as births, deaths, and marriages for the most recent years available. 56 MOBILITY AND TRANSPORT CONNECTIVITY SERIES o Police department: (deputy) police commander and police officers from operational teams in at 3.3. Preparations by the least two regions and from two different levels. In some countries, the police comprise several Review Team entities responsible for urban areas, for the The review team should prepare the visit by countryside, for motorways, and other trans- reviewing the information provided by the host port infrastructure. In that case, it is necessary team, checking the crash data for complete- to meet each of them, or at least the most ness and consistency (if possible), comparing important ones the national figures to those from international sources, and investigating earlier reviews on the o Transport/road/public works department host country and any previously issued recom- o Hospitals or other health centers: staff in mendations. In order to gain maximum informa- charge of registry systems in hospitals or other tion from discussions during the country visit, the health centers where road crash patients are team should carefully review the material in close brought and registered. If this is an important cooperation between visiting and hosting teams. component of data collection, two different The team can also consider conducting a pre-visit hospitals should be visited video conference to elaborate on data needs and o Ministry of Health and staff in charge of man- challenges. aging data from hospitals and health centers The preparations should include the following steps: o Bureau of statistics: staff in charge of fleet and population data, death statistics, vital statistics, • Establish an overview of the organization struc- mobility statistics, infrastructure statistics ture in the host country o Other important stakeholders, such as research • Identify key stakeholders, indicating whom the centers, universities, road safety advocacy, and team should meet during the visit insurance • Search for previous reviews or related research on the host country • Acquire necessary permissions to attend a crash site, together with the investigating officer • Review existing documentation and reports • Inspect crash data: In practice, it might not be possible to accommo- o Compare crash definitions used to United date all visits within the tight schedule of one Nations Economic Commission for Europe mission. The host team will have to do their best (UNECE) definitions to arrange as many of these essential meetings as possible. One-to-one interviews are ideal, but group o Inspect crash data for anomalies meetings could be a valuable alternative—especially o Compare reported crash numbers to World if people feel free to report possible problems. Ide- Health Organization (WHO) statistics ally, the visit should start with the entity in charge of public road safety policy. Meeting this person is the o Compare reported crash data to vital statistics best way to gather a broad overview of the situation • Assess the use of SPI and mobility data from the outset and to pin down the issues at stake. The following sections describe the preparation activ- ities in more detail. 57 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS ESTABLISH AN OVERVIEW OF THE IDENTIFY EARLIER REVIEWS AND OTHER LITERATURE ORGANIZATIONAL STRUCTURE Consultants planning the review should inves- The hosting and visiting teams should work tigate any previous reviews. Often, quite a few together to establish an overview of the organi- reviews have been carried out. Some of them might zation, detailing how different stakeholders work have been data specific and others might have together, the roles and responsibilities at the institu- touched on data issues. It is often informative to tional and staff level, what the crash data collection investigate what has been previously reviewed and chain looks like, and any agreements in place for data what early recommendations have been issued or collaboration. actioned. Before issuing new recommendations, the team should reflect on whether the earlier The hosting team should help the visiting team recommendations have been implemented and, understand relevant processes and procedures. It if not, determine the barriers that impeded their can also be helpful for the visiting team to consider implementation. cultural aspects relevant to the review. Since team members will not necessarily be aware of cultural Also, reviewers should check for other publica- specificities, it could also be useful for the visiting tions of interest about the host country. Interest- team to meet with people who have visited the host ing sources include the Global Status Report on Road country before or have an online meeting with a local Safety 2018 (WHO 2018); the country profiles (World consultant not originally from the host country. Bank 2019), Lancet publications on the global bur- den of disease and injury (for example, Chen et al. IDENTIFY STAKEHOLDERS 2019), or the E-Survey of Road Users’ Attitudes (ESRA) The host team should identify the main actors project country factsheets, available online at https:// and describe their roles in data collection and www.esranet.eu/en/publications/. data usage. During the visit it will be important to look for gaps in engagement with these partners, and REVIEW THE EXISTING DOCUMENTATION AND REPORTS a check of the stakeholders listed by the host team The host team should provide the review team against the list of typical stakeholders (see chapter with prior access to a range of materials, doc- 4) could be useful. Possibly, the team might have to umentation, and relevant studies on the data check with the host team to identify who else could collection systems, such as crash or hospital regis- assist in the collection, analysis, and eventual use tration forms, glossaries, lists of indicators available, of data. As an example, an existing asset database and other materials, along with the outputs based on might prove quite useful, even if not directly linked to road safety data, including reports and dashboards. performance management activity. With translation software continually improving, lan- guage should no longer be a barrier. The review team should also check with the host team whether they have made the necessary The documentation on data procedures can help arrangements to meet people at different levels. To the reviewers prepare for the visit. Moreover, the understand whether general procedures are actually availability of such documentation indicates how well applied in practice, it is important to speak with the these procedures are structured. A glossary for the agency leaders as well as those who attend crash registration of crashes should be included, but, if not, scenes, fill in forms, and perform other tasks related the registration forms for the police and hospitals, if to data collection. available, offer an overview of the collected variables. 58 MOBILITY AND TRANSPORT CONNECTIVITY SERIES Depending on the level of detail, reviewers should or higher, or MAIS 3+ (Weijermars et al. 2018), and check beforehand whether the choice of variables is for slightly injured needing medical help. All crashes adequate, and confirm during the visit whether all involving a moving vehicle—even if it is only one vehi- variables have been filled in reliably. cle and even if this is a nonmotorized vehicle—should be included. The studies and reports provided should indicate whether, and how, different types of data are CHECK DATA FOR ANOMALIES actually used. Progress should be monitored using The review team should have a thorough look at different types of data. The review team should check the crash data before the visit. Individual crash whether all available indicators are indeed evaluated. records, if available, can be checked for complete- Not all analyses are equally meaningful. As an exam- ness and consistency; for example, crashes occurring ple, crash data can be presented simply as numbers, during daylight hours cannot be coded as night-time though it is easier to make sense of the data when crashes; pedestrians should not have been counted rates per population, vehicles, kilometers, and other as passengers, and other potential issues. If no indi- indicators are shown as the development over time vidual records are available, perhaps for privacy or or comparison of percentages in different situations. other reasons, a closer look at the summary statistics If the crash location is available, crashes can (and can also indicate possible problems. For example, the should) be related to road characteristics. Normally, annual data can be plotted over time. Any sudden in-depth reports addressing particular problems or breaks could either be related to major events in the pertinent questions would also be available. During country, such as a pandemic, an economic crisis, or the review, the team should evaluate whether pub- a new law making seatbelts mandatory, or it could lication formats suit the purposes of different stake- point to changes in crash registration. Anomalies, holders. For instance, the police should be able to use such as sudden changes in the development, can be the data to understand when and where enforcement interesting to discuss during the visit. The regional activities are needed, or the data could contain the data can then be compared to check for full spa- necessary network information to inform infrastruc- tial coverage. Typically, one should aim for 5 to 10 ture programs. regions in the analysis and compare, for example, the number of fatalities per population for each region CHECK THE DATA PROVIDED or the distribution of road user types for fatalities in COMPARE DEFINITIONS USED FOR INTERNA- each region, such as the percentage of pedestrians, TIONAL REFERENCE motorcyclists, and car occupants among the fatalities. To compare data in the regional road safety observa- Large differences between regions. For instance, a tories and for benchmarking purposes, it is useful if much smaller proportion of pedestrians among fatali- the definitions applied by the host country agree with ties, can be related to real differences with respect to the definitions applied in international crash statis- traffic in general, such as urban versus rural areas, tics (Eurostat, ITF, and UNECE 2019). For example, the or with respect to road safety such as better infra- criterion for fatalities should be death within 30 days, structure. However, these differences can also point for severely injured it should be 24 hours in hospital, to weaknesses in the data registration methodol- or a Maximum Abbreviated Injury Scale rating of 3 ogy, for example, not recording pedestrian crashes. 59 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS The local experts should be aware of large regional road traffic crash counts. However, the WHO esti- differences, which will either provide an explana- mates can exceed a country’s record by much more. tion in terms of traffic safety, or deduce a possible a On average, middle-income countries (MICs) report data-collection issue—along with, ideally, a possible only 50 percent of fatal crashes, while low-income solution. countries (LICs) report only 10 percent (World Bank 2019). Reviewers should check “Explanatory Note 3” Comparison of fatalities who died on the scene of the Global Status Report (WHO 2018, 288–95) for and who have died in hospital can reveal possi- the methods applied to different groups of countries ble problems with updating crash records when and to which group the host country belongs. For the victim has died in hospital. In European coun- large differences between the WHO estimate and tries, between 17 percent and 43 percent of road the number reported, reviewers should ask the local safety fatalities are reported to have died in hospi- experts for an explanation. Note, however, that often tal (Adminaite et al. 2018). Although differences can countries’ experts do not understand how the WHO also be due to differences in trauma response, lower arrived at the relevant estimates. percentages are likely to point to issues with updat- ing crash records when casualties die in hospital. In Another useful check is to compare the number of Organisation for Economic Co-operation and Devel- fatalities to the country’s vital statistics. If the cause opment (OECD) countries, the ratio between reported of death is included, road crash fatalities can be fatalities and reported hospitalized casualties ranges derived from these. between 2 and 31 (with an average of 9) hospitalized casualties per fatality. However, the true estimated REVIEW REPORTS AND OUTPUTS OF ratio is about 15 hospitalized casualties per fatality REGIONAL OBSERVATORIES (World Bank 2019). The regional road safety observatories should have reports on road safety priorities and data of coun- COMPARE REPORTED NUMBERS TO OTHER STATISTICS tries within a certain region. Comparing the national The World Bank Country Profiles (World Bank 2019) data with other data from similar countries in the share the country-reported data for low- and mid- region would provide insight on a country’s perfor- dle-income countries (LMICs) and contrast them with mance—in data collection in particular, and road the estimated numbers in the Global Status Report safety in general. Outputs from the observatories can on Road Safety 2018 (WHO 2018), along with esti- also provide context as to the most pressing issues in mates based on the global burden of disease (GBD) a certain region, which can then inform on the types data gathered from the Institute for Health Metrics of data to prioritize. and Evaluation’s GBD Results Tool, available on the Global Health Data Exchange: http://ghdx.health- PREPARING FOR THE INTERVIEWS data.org/gbd-results-tool. The WHO estimates are To prepare for the interviews during the visit, at least 3 to 4 percent higher due to differences in questions should be derived from the general definition—for instance, WHO statistics also include introductions in these guidelines and from the deaths after more than 30 days—and because cases issues identified during the preparations. The with “unclear intentionality” are added pro-rata to the materials provided in advance and the details of the 60 MOBILITY AND TRANSPORT CONNECTIVITY SERIES investigation of data completeness and consistency 4 discusses the actors that should be consulted should be reported in the appendix of the review and the topics to examine. A list of questions for report. It can be useful to discuss possible data issues each actor should be prepared and used to conduct identified in these investigations completed during semi-structured interviews. Generic lists of questions the visit. However, the value of triggering interesting that can be adapted to the host country can be found discussions should be balanced with the importance in appendix B. of building trust and getting to know the country. If the visit is mostly about establishing a working Preparing a brief presentation for each meeting relationship, this sort of discussion might better can be useful, which should include the objectives be postponed. If the visit serves to screen the data to be pursued, the role played by the host institution, collection and use, it is useful to have some possible and the main issues to be dealt with in the meeting to issues at hand. One should, however, proceed from help participants understand the importance of the the assumption that a reasonable explanation exists. meeting and why their collaboration is required. The reviewers should be prepared to learn about the local road safety situation and data system rather A brief summary of the preparatory activities is out- than arriving convinced something is wrong. Chapter lined in figure 3.2. Figure 3.2. Preparatory Activities for Road Safety Data Reviews Determine scope and objectives of the review Request relevant data and documentation from the host country Organize meetings ahead of the visit Identify stakeholders and government organization structure Undertake a literature review of published studies and reports Review existing documentation, forms, and reports Inspect crash data Assess SPIs and mobility data Develop a preliminary assessment and insight Prepare interview questions and presentations Source: Original figure produced for this publication. 61 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS References Adminaite, Dovile, Graziella Jost, Henk Stipdonk, and Heather Ward. 2018. An Overview of Road Death Data Collec- tion in the EU. PIN Flash Report 35. Brussels, Belgium: European Transport Safety Council (ETSC). https://etsc.eu/ an-overview-of-road-death-data-collection-in-the-eu-pin-flash-35/. Chen, Simiao, Michael Kuhn, Klaus Prettner, and David E. Bloom. 2019. “The Global Macroeconomic Burden of Road Injuries: Estimates and Projections for 166 Countries.” Lancet Planet Health 3 (9): e390–98. https://doi.org/10.1016/ S2542-5196(19)30170-6. Eurostat, ITF (International Transport Forum), and UNECE (United Nations Economic Commission for Europe). 2019. Glos- sary for Transport Statistics. 5th ed. Luxembourg: Publishing Office of the European Union. https://doi.org/10.2785/675927. Weijermars, Wendy, Niels Bos, Annelies Schoeters, Jean-Christophe Meunier, Nina Nuyttens, Emmanuelle Dupont, Klaus Machata, Robert Bauer, Katherine Perez, Jean-Louis Martin, Heiko Johansson, Ashleigh Filtness, Laurie Brown, and Pete Thomas. 2018. “Serious Road Traffic Injuries in Europe, Lessons from the EU Research Project SafetyCube.” Transporta- tion Research Record 2672 (32): 1–9. https://doi.org/10.1177/0361198118758055. WHO (World Health Organization). 2018. Global Status Report on Road Safety 2018. Geneva: World Health Organization. License: CC BYNC-SA 3.0 IGO. https://apps.who.int/iris/bitstream/handle/10665/276462/9789241565684-eng.pdf. World Bank. 2019. Guide for Road Safety Opportunities and Challenges: Low- and Middle-Income Countries Country Profiles. Global Road Safety Facility Report. Washington, DC: World Bank. https://www.roadsafetyfacility.org/publications/ guide-road-safety-opportunities-and-challenges-low-and-middle-income-country-profiles. 4. Country Visit: Stakeholders to Meet 63 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS All entities involved in the collection or use of road safety data should be consulted. This concerns 4.1. Ministries and governmental actors, the institutions that deal with crashes directly, as well as other stakeholders. Government Departments During the visit, the most important stakeholders should be consulted on all issues relevant to their The ministry responsible for road safety is often the role in the collection and use of the different types transport department. Even when this is not the of road safety data. case, the transport ministry should be using road crash data to define blackspot locations and deter- This chapter discusses the entities typically mine road design. These departments also com- involved in a road safety data review. Topics that monly collect data related to driver licensing and should be discussed during a review often overlap vehicle registration. between different stakeholders, because one of the important objectives of the data review is to address In addition, road safety should also be the responsi- inconsistencies between the approaches of differ- bility of the health departments, and where possible, ent institutions. The essential topics are listed in the the review team should identify the specific depart- chapter 5, with indications of which stakeholders ment(s) tasked with collecting crash data. This might should typically be consulted on each topic. be the public health or the epidemiology department or could be the rehabilitation, emergency, or post- The review team should attempt to meet with crash department. Different departments often col- all stakeholders, including ministries involved in lect specific types of crash data. road safety, police, statistics offices, road safety advocacy, and road safety scientists. For those In some countries, the ministry of the interior is bodies who report the data, the review team should responsible for road safety or can provide data on meet staff at various levels, such as local, regional, police interventions. national, management, and those executing the work. Consequently, the country visit will likely be The lead road safety agency is usually situated in one densely packed with interviews. of these ministries. It should include a group of data analysts charged with aggregating all data, including The procedures might not be uniform through- the questions stakeholders have, the available data, out the entire country, and could differ for states, and corrections for possible problems with the data. provinces, or even municipalities. For instance, each This group serves as the natural hosting team for the police district and type of hospital might have differ- data review. These analysts should benefit most from ent protocols. In some countries, various police enti- the review in terms of capacity building, but should ties oversee different parts of the road network. In all already have the capacity to carry out meaningful these cases, speaking with representatives from each analyses. The group’s output should reflect the needs active institution should be prioritized. and expectations of policy makers and road safety 64 MOBILITY AND TRANSPORT CONNECTIVITY SERIES advocacy groups. The analysts are also responsible for making the data available to others. For their 4.2. Police analyses, they should be aware of how the data col- lection process functions. They should check the data The review team should assess the police’s com- and—if they become aware of problems—would be mitment to collecting crash and other road safety responsible for initiating work to correct them. data, to the practical implementation of data collection, and also their use of the data. Review- Ideally, the topic of road safety should be ers should inquire how well the police understand addressed cross-sectionally across relevant orga- the reason these data are collected, and the impor- nizations. An interministerial or interagency meeting tance of this task in managing road safety outcomes. should be organized to ensure statements from one If the police actually use the data they themselves agency are validated or cross-checked with state- collect, this tends to increase their commitment. If ments from another agency. The review team should different police bodies are involved in data collection, also meet with the departments not yet involved in they should ideally all feed into the same national road safety work to check for additional data sources database and use the same (or sufficiently similar) and gauge whether policy makers might support and data collection protocol. With separate databases, all request additional data collection, and if so, whether issues concerning data transfer must be investigated. they could provide the necessary resources. The Moreover, it is essential to speak with personnel in departments should also be asked about their use of each of the police bodies and examine whether the road safety data. procedures are sufficiently similar. At least two sep- arate police departments should be visited. Ideally, Subnational authorities can play an important role in one of them should be located in the capital city, or road safety data collection especially if the relevant wherever the review is situated, and serve as police competencies, such as infrastructure, policing, or headquarters or other office high up in the hierar- laws, are not situated at the federal level. In that case, chy. The team should also visit a police department speaking with subnational representatives separately, in a contrasting region, possibly a rural or more or bringing them together for a group discussion, can distant office, and of a lower hierarchical level, per- be very helpful. While these guidelines address a data haps a regional office. The interviews should not be review at the national level, this type of review could restricted to the management, but also include field also inform local or subnational authorities looking to officers who attend crash scenes. If any other per- assess their data management practices. The applica- sonnel enter data into crash forms, they should be bility is, however, limited to the type of data for which interviewed as well. the authority in question is responsible. 65 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS 4.3. Hospitals 4.5. Coroners Mortality statistics can be used to check crash Vital registrations of deaths are collected from data, or might even serve as their base. The data data provided by doctors, hospitals, and coro- reported to the World Health Organization (WHO) ners. In most countries, violent deaths, such as road contain a code for cause of death, which identifies crashes, require forensic action, and all three groups casualties of transport accidents. However, those would, understandably, appreciate knowing about statistics are also not necessarily complete. Where working procedures relating to road crashes. This hospitals play a vital role in the registration of road interest applies especially to coroners, who carry crashes, the reviewers should investigate how reg- out analyses of psychoactive substances on corpses. istration works in practice. This could concern the Results of the coroners’ procedures can be used to registration of details about the crash in the hospi- check and augment the road crash database. The tal’s database, such as the travel mode of the victim reviewers should investigate how coroners collect and the opponent. Cooperation with the police to data, how they classify road traffic fatalities, and improve crash records is also important, including whether they report to the road traffic accidents notifying the police of the crash, determining the database or to the traffic police. severity of resulting injuries, and updating the crash report when needed, for example, when a patient dies. A third topic of interest concerns collecting data 4.6. Insurance Companies on trauma response, such as how long an injured casualty receives medical treatment after a crash. Although they do not function in an official capac- ity, insurance companies may play a significant 4.4. Statistics Office role when constructing traffic-related indicators. For instance, the Dominican Republic produces road fatality numbers by aggregating the databases from In many countries the national statistics office law enforcement officers, forensics, hospitals, and collects data from the police and hospitals. This vehicle insurance companies. A similar approach is office might be the entity responsible for checking, being planned in Thailand, where the insurance com- correcting, and augmenting the data, such as by panies hold the most comprehensive crash datasets linking them to other data sources. The statistics in the country. office might also maintain other relevant databases, including vehicles, drivers’ or riders’ licenses, fines, and verdicts for offenders, and also maintain the vital statistics used to check the number of fatalities. 66 MOBILITY AND TRANSPORT CONNECTIVITY SERIES 4.7. Road Safety Advocacy 4.8. Research and Academia Groups and Journalists Research and academia play an important role in road safety policy-making efforts. The lead agency Advocacy groups are important users of road should be supported by analytical work conducted by safety data. Reviewers could benefit from interview- academic researchers, making sure they have access ing these groups. Suggested questions include the to all road safety data. In additional, researchers can following: Do they have access to data?; and Do they often help with technical issues, such as matching use road safety data? Moreover, advocacy groups hospital data with police data, the weighting of road- often collect data as well. Are the data shared with side survey results, or location mapping and spatial other actors? In addition to actual lobby groups, modeling of crashes. journalists can also be important users of road safety data. They can play a role in increasing awareness of the need to improve road safety, but they can also contribute to a hostile climate towards road safety interventions and data collection activities. 5. Interviews: Topics to Address 68 MOBILITY AND TRANSPORT CONNECTIVITY SERIES In the interviews, reviewers should focus on prac- members of different organizations is also important. tical information and collecting evidence of how They should be aware of each other’s actual proce- procedures are actually carried out, as compared dures in data collection and coordinate their efforts, with the official protocol. It is important to talk to for example, police and health bodies. Consequently, those persons who implement the procedures, such the same questions should be put to many actors and as officers at the scene, those who enter the data, the consistency of their responses checked. and others. Process demonstrations and answers to practical questions can help reveal structural prob- Meetings with stakeholders should be conducted lems. This chapter suggests questions for the vari- as semi-structured interviews. A generic set of ous relevant topics. interview questions for the different stakeholders can be found in appendix B. To help prepare for the The reviewers should be prepared for differing interviews, this set of questions should be adapted to cultural contexts influencing behavior, such as the situation in the country under review. During the interviewees being hesitant to answer any question interviews, the reviewers should use the questions as with “no” or other negative response, or admit to a a starting point and as a checklist to validate whether mistake or being involved in corruption. Such aspects all relevant topics have been discussed. Following the can not only affect data collection practices, but also results of the first interviews, the questions for later what interviewees say and how they say it. Obtaining ones might need to be updated. Group discussions concrete, accurate information requires the review can be a time-efficient way to get input from several team to focus on practical implementation, which people, although some people may be uncomfort- helps in evidencing the procedures as properly car- able discussing problems in this setting. Sometimes ried out, as compared with the official protocol. one-on-one discussions will encourage more open responses. Practical demonstrations and observing the exe- cution of different tasks offer a more accurate The topics included in the interviews should picture than verbal explanations. For a more vivid, cover the whole chain of investigation, regis- complete picture of the country’s road safety data tration, transmission, storage, processing, and situation, the review team should ask to visit a crash use of data. The topics covered here concern the site with the police, or visit a hospital or health cen- police crash data, the hospital injury data, and other ter. They should always ask to see concrete examples road safety data, such as safety performance indi- and materials. Focusing on practical restrictions in cators (SPIs) and travel data. Each of the topics in terms of time, transport, and equipment can help the the remainder of this chapter should be discussed review team assess whether the outlined procedures with different actors to find out not only how things are truly being implemented. are supposed to occur, but also how they are imple- mented in practice. An important objective for Reviewers should gauge whether all stakeholders reviewers is to make sure the practices of different understand and support the need for quality data. actors converge. The following sections provide for Moreover, commitment to identifying and correct- each topic concrete instructions on which aspects ing possible problems is a necessary, but not always the reviewers should consider and which ques- given, condition for progress. The interaction between tions should be asked. 69 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS 5.1. Organization of Crash of reported crashes, the types of vehicles involved, the time required for registration, and a quick Data Collection assessment of whether it all adds up can help indi- cate circumstances in which crash scenes cannot be by the Police attended. In cases where the police cannot attend the crash scene, are crash data collected in other ways? Everybody who interacts with police data should be aware of how they are collected. Accordingly, INFORMATION INCLUDED IN CRASH DATA the questions below should be put to both the police What is reported about crashes and how? Which as well as analysts in the lead road safety agency and variables are coded? If the registration form has not the statistics office, if they handle these data. been provided in advance of the visit, the review team should inquire and attain a copy. Is the form NOTIFICATION online or on paper? Is the form uniform throughout How are the police notified? Is there a central the country? Which tools are used during the crash emergency number? Who notifies the police about investigation, such as a measuring wheel or laser crashes when victims are taken to hospital in private device, alcometer, global positioning system (GPS) cars? Could there be crashes of which the police are device, video camera, and other tools? Can these be not notified, for example, crashes involving only one shown or demonstrated to the review team? vehicle or only nonmotorized vehicles? What happens if participants are not insured—do they still call the For several variables, reviewers should discuss police? What are the possible reasons why people how each is coded and, accordingly, how the offi- would not report a crash to the police, such as driving cer at the scene would correctly determine the without a license, driving while intoxicated, or having coding. Chronically difficult issues include, for exam- a general distrust of police? ple: use of protective equipment, intoxication, speed- ing, and injury severity. Is the hospital contacted about REGISTRATION OF CRASHES whether a casualty’s condition has declined or, particu- Do the police have the budget and personnel larly, if a casualty has died in hospital? If so, is the infor- resources to attend every crash scene, at least mation on injury severity updated in the database? those resulting in serious or fatal injuries? What about remote areas, geographic or climate condi- If causation factors are coded, reviewers should tions, or too many crashes? Could these circum- check on which information these factors are stances result in not attending a crash scene? Are based. Such factors are generally recorded without enough patrol vehicles available to attend every crash the officers being able to obtain reliable information. scene? Reviewers should ask very specific questions Moreover, the factors available for selection often about all practical aspects of how police process crash focus mostly on driver errors, rather than addressing data, such as the number and length of work shifts, all aspects of the road traffic system, including road how many officers attend a crash scene, the number user behavior, infrastructure, and vehicles. 70 MOBILITY AND TRANSPORT CONNECTIVITY SERIES DATA TRANSMISSION LEGAL FRAMEWORK When and where is the crash entered into the Which crashes are the local police legally obliged database? Directly at the scene? Afterwards? How to report: all, all injury crashes, all fatal crashes? long after attending the crash scene? Is the person Are the police obliged by law to share the data? Or who fills out the crash registration form the same as do any legal constraints, for example, privacy restric- the person attending the scene? If not, which infor- tions, prohibit this? Does the legal framework identify mation is the entry based on and is the completed other stakeholders in crash data collection? registration form checked at the local police station? CHECKING DATA AND FEEDBACK TO LOCAL POLICE If the initial registration is on paper, who is respon- Are there checks or controls on whether the sible for entering data into the database? Is the police fully report on all the crashes they are registration form sent to a central location? Is some- obliged to record? How? Are there consequences one at the local police station checking whether all for the local police if not all crashes are reported? Are registration forms are completed and sent to and then quality checks conducted on crash reporting? Is the imported into the central database? What process completeness of the variables filled in checked? Are ensures each form is counted? Is double counting pos- consistency and plausibility checks routinely, or auto- sible? And if so, what is done to avoid it? What about matically, conducted? Are the results fed back to the data updates—for example, a crash that turns out to local police? To the officer who coded the case? be fatal three weeks later? How long does it take for data to reach the crash database following a crash? What is, or could be, done to motivate the police regarding data registration? Is the registration of HUMAN RESOURCES any value to the local police? Do they get reimbursed? What kind of training do officers who investigate Are database entries also used to produce reports for crashes receive? How many days? By whom? Are all the court? Do the local police use the summary infor- officers trained? Are they all trained in the same way? mation on crashes themselves, or are they aware of Are any follow-up or refresher courses offered? Does any local policy makers using the data? the training succeed in highlighting why recording crashes is so important? If data analysis is an officer’s duty, is this addressed in training? Is there sufficient budget allocated for training? Are there training standards that cover how to process and analyze crash data? 71 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS 5.2. Surveillance of Road Traffic Injuries by Hospitals Hospital data are an important complement to Is the system set up exclusively for the collec- data collected by the police. If hospital data are tion of data on road traffic casualties? If not, how used for statistics on road traffic casualties, the local can road traffic casualties be selected or disaggre- staff involved in registration should be interviewed as gated from other forms of trauma? How is whether well as the management of the hospitals and institu- a patient is a road traffic casualty or not determined? tions that further collect and process the data. Ques- What is the definition of a road traffic casualty? tions that should be considered are described in the Which variables or characteristics are included in following paragraphs. the injury surveillance system? How does the system work in practice? When is a case reported—directly, MEDICAL TREATMENT OF ROAD TRAFFIC CASUALTIES or based on information from another system or Which hospitals or health facilities treat road traf- registry? Who enters the information in the injury fic casualties? Are all road traffic casualties treated surveillance system? Are these people trained to do in hospitals? What if the casualty is not insured? this? How long and how thorough is the training? Is How do road traffic casualties arrive at the hospital? follow-up or refresher training offered? Do they have Mainly or all by ambulance? Are there also private adequate time to enter the data? ambulance services? What happens after the casu- alties arrive—first the emergency department, then ASSESSMENT OF INJURY SEVERITY admission to the hospital if needed? Which other How is a serious injury defined? Are they familiar data sources could be used—emergency calls, ambu- with the MAIS scale? The Abbreviated Injury Scale (AIS) lance dispatching, insurance? Could these be used to measures the severity of each injury on a scale from 1 check the completeness of the casualty registration? (slight injury) to 6 (non-treatable, usually fatal injury) Or to address post-trauma care? Or to investigate the for each of the 9 regions of the body (head, face, medical costs related to road crashes? neck, thorax, abdomen, spine, upper extremity, lower extremity, external, other). The Maximum AIS (MAIS) INFORMATION INCLUDED IN ROAD determines the highest AIS-score out of all injuries INJURY SURVEILLANCE SYSTEM sustained (AAAM 2016). In many countries. MAIS 3 or Is there a system for registering road traffic inju- higher is considered a serious injury (Weijermars et ries? How do hospitals collect and report the data? al. 2014). Do they include the AIS information in their How is each field in the hospital record form filled in? reports or intend to do so? If not, what method is used Which information about the crash is included? Road to determine whether an injury is considered serious? user type of patient? Road user type of the other involved party, if any? Information about the loca- DATA TRANSMISSION tion (urban, rural, type of road)? What about data for Are procedures and forms uniform throughout patients transferred to another hospital? Would the the country? How is data from different hospitals established cause of the injuries be included in the combined? What happens to data for patients who records of the second hospital? are transferred to another hospital; could they be double counted? Which casualties are recorded in the system—only people admitted to hospital, or outpa- tients as well? 72 MOBILITY AND TRANSPORT CONNECTIVITY SERIES HUMAN RESOURCES Who enters the information in the injury surveil- 5.3. Storage, Integration, lance system? Does the staff have adequate time to do this? Are they trained to do it? How long is the and Quality Control training? Is follow-up or refresher training offered? The use of road safety data depends crucially on Is there sufficient budget allocated for training? Are how it is integrated, stored, and made accessible there training standards? Does the training include to all stakeholders. Different data sources must be information about road crashes and the importance linked to ensure completeness and applicability and of injury surveillance to improve road safety? the data quality should be controlled and fed back to those who collect and enter the initial data. The follow- LEGAL FRAMEWORK ing questions help to assess these important steps. Are all hospitals and health centers obliged to register casualties in the injury surveillance sys- DATABASE AND ACCESSIBILITY tem? What happens if they fail to do this? Are there The existing databases, for crash or hospital data, incentives for the hospital to register road traffic for example, should be assessed by means of casualties? Do private hospitals report as well? Do practical demonstration to understand the features any privacy restrictions prohibit reporting of road included and see whether the appropriate people traffic injuries? Is hospital staff obliged to inform the know how it is used. Additionally, the following ques- police about incoming road crash casualties? tions could be asked: CHECKING DATA AND FEEDBACK TO LOCAL HOSPITALS Who is maintaining the central crash and casu- Are the data checked for missing records, com- alty database? Does it combine different sources, pleteness of records, or inconsistencies? If so, how? such as police, hospital, coroner data, or civil registry? Is feedback on data quality provided to the hospitals What is the architecture of the crash database? Is the and local staff? crash information linked to other databases, such as the vehicle or license registry or asset databases for Are there any indications about the degree of road management? Does the coding of data allow for completeness or about under-registration? Are cross-referencing between datasets? there other databases from emergency rooms, ambulance dispatchers, and other sources? Could Who feeds the database? Who is authorized to this information be useful in checking the registration make changes? Who has access? Is there a data of road traffic casualties? warehouse that enables the combining of crash and casualty data with mobility data or SPIs? Is the infor- mation aggregated or linked? Are there any standard outputs (maps, dashboards, visualizations, or stan- dard reports) produced automatically? 73 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS INTEGRATING INFORMATION FROM DIFFERENT BODIES problems? Missing cases? Inconsistent coding? Many Hospitals, police, and other actors, such as the incomplete fields? Are the data corrected, and if so, civil registry or coroners, should cooperate to how? Are the results fed back to the collecting bodies, ensure the identification and correct classifica- that is, the police and hospitals? tion of all road traffic casualties. It is important to identify gaps in the chain of registration so these Are estimates of under-registration included? can be corrected later by aggregating different How is under-registration estimated? If discrepancies databases. appear with other figures, such as those in the Global Status Report (WHO 2018), are the local actors aware Do medical staff check whether police are informed of that? Do they have an explanation? when they treat crash casualties? Do the police consult medical staff in reporting the injury sever- The reviewers should ask to see individual crash ity? How are changes in injury status handled? Most records. These should be checked for completeness importantly, what happens if a crash victim dies in and consistency. If the preparatory analysis, including hospital? Who takes the initiative to update the crash differences across regions, has indicated any abrupt record? Do police contact the hospital after 30 days changes over time or large differences between to check for any update to the recorded severity? Or regions, the review team should discuss any possi- does the hospital notify the police if a crash victim ble explanations in terms of “real” differences with dies in hospital? What if a casualty dies somewhere respect to factors such as traffic participation or road else? For example, in some cultures, families bring structure, or whether the reported differences point the patient back home before he or she passes away. to inconsistencies in registration practice. Would these cases be identified in the civil registry or by a coroner? COMPLETENESS AND QUALITY OF DATA 5.4. Other Road Safety Data The road safety agency, the statistics depart- ment, or central police management, civil regis- Road safety data includes crash data as well as try, and road authority could (and should) all con- data about the safety performance of the road tribute to the various integrated data sources at traffic system and interventions to improve road central level and help enhance data quality. The safety. These data are best used when combined following questions should be considered: with other data such as traffic data, such as distances traveled and the split between different transport Are the data checked at central level? Are the modalities. While the focus of a limited review should numbers comparable across datasets? Vital statis- be on crash data, the availability and use of other tics? Coroner data? Are police data and hospital data road safety data should be discussed, along with the matched, or at least the numbers compared? possibility and methods of collecting other road safety data, with the department of transport, police, hospi- Are quality checks conducted to ensure consistency tals, statistics office, and road safety researchers. and completeness? What happens in the case of 74 MOBILITY AND TRANSPORT CONNECTIVITY SERIES MOBILITY DATA SAFETY PERFORMANCE INDICATORS Travel data are important in interpreting crash As intermediate outcomes, SPIs are essential in and casualty data in relation to the distances linking countermeasures to final outcomes (crash traveled or with surrogate measures of them. and casualty data). The review team should discuss The review team should explore what kinds of data the concept of performance indicators with different are available, such as distance traveled, population, actors in the lead agency as well as in other ministries vehicle registration, road length, and gross domestic or road safety advocacy groups. If appropriate, their product (GDP). value in monitoring and evaluating the effect of road safety measures should be explained. The following For dedicated data collections on mobility, the team questions could be considered: should check the following issues: • Which SPIs are available? Are evaluations of road • Which travel modes can be differentiated? user behavior (speeding, use of protective equip- ment, intoxicated driving, distractions) available? • At what frequency is the data collected (perma- Could big data sources, such as mean speed data, nently, annually, ad hoc, or other) help? Are roadside surveys conducted? Are data • Are the data representative (see appendix C)? on underlying attitudes available? Are data on the o What is the spatial resolution for which you can safety scores of roads or vehicles available? Is an make meaningful statements (country, region, asset database for the maintenance of the road municipality, road)? network available? For the available SPIs, ask for details on data collection—including representa- o What is the temporary resolution for which you tiveness, observed versus self-reported, and sam- can make meaningful statements (year, month, ple size—and be sure to inquire about the use of day, hour, or other)? the resulting figures. • How are these data used? • Which SPIs would be considered the most useful for the review country? Which would be the most If no dedicated data collection on mobility is yet in achievable to measure? For example, the use of place, discuss what could be done. Explore the use of protective systems is relatively easy to observe “smart” measurements, such as smartphone apps or systematically, while even countries with long road data from telephone companies. However, it might safety research traditions struggle to measure dis- very well be the case that mobility questionnaires are traction or to produce safety scores for their road still the cheapest way of establishing the share of dif- networks. ferent transport modes in traffic. To relate the devel- opment of casualty numbers to the development in Many SPIs require roadside surveys (see appendix C). motorization, fleet size by vehicle class, fuel sales, and However, while observed behavior is more reliable, even GDP can be interesting surrogates. self-reported behavior can be a much cheaper alter- native, because several behaviors can be addressed at once (see the section on Road user behavior in chapter 2). To measure speeding, floating car data 75 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS provide an alternative to classic measurements with issue? Could these barriers be overcome? Which radars or tubes. Also, big data companies share agreements could be made to deal with data pro- average speed data per road segment, which can be tection issues? Could intersectoral commitment matched with crash data (see the Infrastructure sec- be increased? Which level would, in that case, be tion in chapter 2). the most important to address? Reviewers should discuss with the department of • For the road administration, indicators could transport which indicator could be collected to mon- include stretches of road with reduced speed itor physical infrastructure improvements (see the limits, length of road sections, number of intersec- Infrastructure section). Consideration could be given tions, or the resources spent on road maintenance to road-related SPIs, such as monitoring the Interna- and redesign. An important question here is tional Road Assessment Programme (iRAP) star rating whether these data are systematically registered. of road sections or implementation indicators, such Is this the case for all parts of the country? Who as the number of improved intersections. would take the initiative to combine data from dif- ferent (for instance, regional) offices? Are all actors IMPLEMENTATION INDICATORS willing to make public how they spend their time To keep track of the implementation of and money? countermeasures and to evaluate their effect, it is important to collect data on all types of • Is vehicle testing mandatory? Which institution intervention. is responsible for summoning owners to test their vehicles? Are data available from testing centers, For example, police data could include the number of concerning the number of vehicles checked, the checks conducted, the number of camera readouts, number of vehicles that complied or failed the the number of tickets and fines imposed, the degree safety tests, and the percentage of vehicles admit- to which controls and distribution of fines are autom- ted after repairs? What is the frequency of test- atized, or the time passing between offense and ing, is it tied to annual or biannual registrations? sanction. Data collection for these indicators is often Which specific safety elements are tested? impaired by practical or institutional barriers. Con- sider the following questions: • The emergency services are important sources for data on post-trauma care. How are ambulances • Do the police register their working hours accord- dispatched? Do these services maintain data on ing to particular activities, for example, alcohol the number of deployments and their timings? checks, speeding controls, and other activities? Would these data allow estimated percentages of Is this done by all police entities? Can these data crash scenes attended by an ambulance and the be aggregated? Is the same software used by the time needed for the ambulance to get there? various entities? Are there institutional barriers to releasing these data, such as in some countries For more information on this subject, see Bliss and where publishing internal procedures is a sensitive Breen (2013). 76 MOBILITY AND TRANSPORT CONNECTIVITY SERIES 5.5. Capacity The review team should evaluate the capacity background? How are they trained for this task? for all functions in the data collection and anal- Do they know anything about road safety? How ysis process. Capacity can concern knowledge, much time is needed to enter each case into the time, or materials. The following questions could be system? Do the staff have adequate time to com- useful: plete this task? Are means (such as software or tools) available to ease the burden of data collec- • How is a sufficient level of knowledge ensured? tion or registration? What are the backgrounds of the employees? How much time do they typically spend on the Storage and analysis: job? Is there a problem with turnover? How long do employees, on average, stay in that position? If • Maintaining a crash database requires techni- on-the-job training is offered, how does that work cal resources and know-how. What is the archi- in practice? Are seniors systematically paired with tecture of the system? Is it a relational database? juniors? How are the rules and methods passed Is the database linked to other databases? Is the on? If formal training is provided, is everyone hardware sufficient to support this structure, and trained? If not, are the contents actually passed any possible improvements? Are the persons who on to other colleagues, and if so, how? If training maintain the database the same as those who programs are offered, the review team should talk built it? Are they data scientists? If not, what is to those responsible for organizing the training their level of expertise? as well as those who should, presumably, have • Does the system allow all necessary analysis? received it. Does it serve the needs of all stakeholders? Can the data be cross-tabulated, for example, can an Moreover, the availability of the necessary analysis be conducted to determine number and resources in terms of materials and time foreseen location of child pedestrians walking at or around for the task should be checked. Specifically, the school start times? Are spatial analysis tools or following questions could be considered regarding tools available to determine crash risk for different specific tasks: areas or parts of the network? Data collection: • Who are the people making use of the data? What is their background? Is training provided for • What capacity is there to investigate crashes them? For how long have they been on the job? at the scene? How are the officers who attend Which software and analysis techniques do they the crash scenes trained? Do they enter the crash typically use? Do they have the capacity to run data into the system themselves or does someone analyses meaningful to stakeholders? As a mini- else do this? If other people do it, how are they mum, these data users would need to know how trained? Is adequate time provided to attend the to query the database, how to cross different vari- crash scene? Is the necessary material available? ables to identify relevant subgroups of cases, and Is the software sufficient to enter the records into they should also be able to master pivot tables. If the database? Is work time allotted specifically for these users exist, the review team should check if data entry? they have identified any projects or tasks where • In hospitals, who registers road crash vic- the methodology or software is lacking. tims on a database? Do they have a medical 77 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS References AAAM (American Association for Automotive Medicine). 2016. The Abbreviated Injury Scale © 2005. Update 2008, edited by T. Gennarelli, and E. Woodzin. Chicago, IL: AAAM. https://www.aaam.org/abbreviated-injury-scale-ais/. Bliss, Tony, and Jeanne Breen. 2013. “Road Safety Management Capacity Reviews and Safe System Projects Guide- lines.” Updated edition. Global Road Safety Facility, Washington, D.C. http://documents.worldbank.org/curated/ en/400301468337261166. Weijermars, Wendy, Niels Bos, Annelies Schoeters, Jean-Christophe Meunier, Nina Nuyttens, Emmanuelle Dupont, Klaus Machata, Robert Bauer, Katherine Perez, Jean-Louis Martin, Heiko Johansson, Ashleigh Filtness, Laurie Brown, and Pete Thomas. 2018. “Serious Road Traffic Injuries in Europe, Lessons from the EU Research Project SafetyCube.” Transporta- tion Research Record 2672 (32): 1–9. https://doi.org/10.1177/0361198118758055. WHO (World Health Organization). 2018. Global Status Report on Road Safety 2018. Geneva: World Health Organization. License: CC BYNC-SA 3.0 IGO. https://apps.who.int/iris/bitstream/handle/10665/276462/9789241565684-eng.pdf. 6. The Review Report 79 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS The review report should describe the context of » Data collection process the review and identify the key stakeholders and – Notification their role in data collection and use. The report should then describe the data systems reviewed in – Recording at the crash scene terms of content, data collection procedures, stor- – The path of the data age, and use. The integration of the various data – Checking and corrections systems should be discussed. Recommendations should consider organization, methods, training, and » Database: storage, availability, and provision communication. of data – Architecture Although the structure of the report depends on the scope defined and on the data structure in – Linking to other types of data the country, a general framework for the review – Accessibility and functionality report, such as the following recommended struc- » Staff, budget, equipment, and training of the ture, can be helpful: actors involved • Context » Evaluation of crash and casualty data o General description of (road safety) situation – Completeness o Earlier reviews, past research – Quality o Context of the present review – Uniformity • Key stakeholders in the road safety data collec- o Other road safety data tion system (collectors and users) » Mobility data available o Levels of understanding of importance of road safety data by key stakeholders » Road safety performance indicators available o Capacity in collecting, processing, analyzing, and reporting road safety data » Quality of each data type • Safety data currently collected – Representativeness and availability o Crash and casualty data (by police and/or – Compatibility with crash data hospital) » Links between road crash database and » Crash and casualty data available other data: frequency, automation, and scale (at the crash, local, regional, national, – Definitions or global level) – Variables • Evaluation of road safety data completeness – Time series (compare with the Checklist on Monitoring and Evaluation in appendix A) 80 MOBILITY AND TRANSPORT CONNECTIVITY SERIES • Use of road safety data • Appendices o Accessibility o Data analyses o Analyses (statistical and thematic relevance, » Comparison with internationally reported consistency, and consideration of time series) casualty figures o Relation to policy, policing, and engineering » Consistency across regions o Combination of different data sources o Short report on each visit o Integration in pre- and post-evaluation of infra- In most cases, crash and casualty data collection structure or development projects can be based on police data, on hospital data, or • Summary of observations on both. The structure when describing the data o Completeness, quality, and uniformity of data collection system(s) depends on the relative impor- collected tance of these systems and on the stage at which these data are aggregated, if at all. In most cases, o Use (and combination) of all data presently the topics listed under the bullet, “Crash and casualty collected data (by police and/or hospital),” will be covered first in o What do the actors themselves want to relation to police data, then the same list of topics will change? be discussed in relation to hospital data, and finally, there should be discussion of whether and how hos- o Observed strengths and possible pital data are used to check and extend the police- improvements based crash data. However, if the reporting system is • Reference to earlier reviews, if applicable strongly based on hospital data, or if the two sources o Implementation of previous recommendations are aggregated early on, the structure might look somewhat different. The evaluation section for crash o Barriers to implementation and casualty data should address all data sources • Conclusion with (new) recommendations and a used. If police and hospital data are combined, the road map for implementation review team should evaluate this combination. o Organization In the evaluation and recommendation sections, o Method organizations, methods, training, communication, o Training and additional data needs should be addressed. For organizations, their structure and cooperation o Communication with different institutions involved should be eval- o Suggestions for data to be collected uated, including the number of “stations” that data 81 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS have to pass before they are aggregated in a data- design and maintenance, for police to target enforce- base, or different practices for crash investigations in ment activities, and address budget concerns—or different police bodies. An important point is whether using data to raise awareness about the conse- the actors have the necessary resources available, quences of road crashes and priority risk factors. such as budget, personnel, hardware, software, vehi- cles, and others. For the various stakeholders the Before suggesting additional data needs, the rec- review team should assess whether they are commit- ommendations should first focus on the use and ted to the need for road safety data, and if not, how combination of all existing data sources. Data are they could be convinced, for instance, by which kind collected, but not yet used for road safety (for exam- of output. ple, vital statistics and coroners’ data, vehicle registra- tion data, and road management data) should receive The evaluation of methods should, for example, special attention, including recommendations for and look at possible improvements in crash investi- benefits of using these data in road safety analyses. If gations, including variables recorded, how causal opportunities and needs for collecting additional data factors are identified, the way data are checked and are identified during the meetings, this should also linked to other data, the structure and maintenance be included. of the database; or the data analysis options pro- vided by the database. Moreover, suggestions could Generally, the evaluation should focus on the be made for meaningful comparisons not yet made strong points first and also report what the or for types of output that would be meaningful for actors themselves want to change. Suggestions those who feed the database and other stakeholders. for improvement should indicate the priority for each recommendation and take into account the efforts In terms of training, a description should be given and resources needed to implement them. A good of which training needs have been identified and how way of presenting proposals is to position the actions these could be addressed in the short term. Changes on a diagram according to their degree of difficulty that would ensure capacity in the long term could be on the one hand, and their degree of effectiveness on suggested. Important topics would include training the other. In addition, a budget estimate should also continuity and the training of new people. be developed as well as a work plan for data improve- ment programs. Finally, it is particularly important to With respect to communication, recommendations provide sound reasoning why these efforts are nec- could be made on how data can be used and pre- essary and what advantages will be gained by their sented in an accessible format. Suggestions could implementation. include products tailored to stakeholders’ specific needs—such as for road engineers to improve road 7. Conclusions 83 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS Road safety data are important to support evi- should therefore be tailored to its needs, referring to dence-based decision making in a country. They international standards for the sake of comparability help in understanding the social and economic costs of the data can be useful—allowing the review team on the issues at hand, selecting effective counter- to benchmark and identify countries with similar measures, setting targets, and monitoring progress. problems and good practices that could help solve Many countries are building up their capacity for data specific issues. With this in mind, these guidelines collection and use, and the World Bank intends to give an overview of the international standards for support this process. casualty data, but also of SPIs and mobility data. A data review can help countries in understand- The guidelines are further meant to provide ing the importance of road safety data, pinpoint- support to road safety experts who will review a ing possible sources of under-reporting of casual- country’s road safety data—be they local or inter- ties, help make maximal use of all available data, national experts. The various steps in the review and possibly, identify additional data and indica- process examine the whole data collection chain, tors. The data review should help to build up capacity data accessibility, its use, and the engagement of in the host country and will often be conducted by different stakeholders in supporting these activities. a joint team of local as well as international experts. In this way, the guidelines will support the work of It can serve to screen the data collection process regional road safety observatories. and identify problems with using the data. Ideally, it would also set up a partnership arrangement or men- The data review described here will be the first toring relationship that could be continued online step in a process to improve road safety data after the visit to follow up on issues identified during management. A sample terms of reference for pro- the visit. curing consultancy services to lead the data review is attached as appendix E. Following the data review, Very importantly, road safety data are much more recommendations should be developed that include than just casualty data. They also comprise safety prioritizing actions, identifying resources, suggesting performance indicators (SPIs), contextual data, such a workplan to move forward, and convincing policy as traffic data, and implementation indicators. While makers to further support road safety. each country is different and data collection and use 84 MOBILITY AND TRANSPORT CONNECTIVITY SERIES Annex A. Sample Monitoring and Evaluation Checklist CHECKLIST: MONITORING AND EVALUATION Questions Yes Partial Pending No For each category of roads (national, regional, provincial, city) are sustainable systems in place to collect and manage data on road crashes, fatality and injury outcomes, and all related road environment/vehicle/road user factors to achieve the desired focus on results? For each category of roads (national, regional, provincial, city) are sustainable systems in place to collect and manage data on road network traffic, vehicle speeds, safety belt and helmet wearing rates to achieve the desired focus on results? For each category of roads (national, regional, provincial, city) are regular safety rating surveys undertaken to quality assure adherence to specified safety standards and rules, to achieve the desired focus on results? Risk ratings? Road protection scores? For each category of roads (national, regional, provincial, city) are systems in place to collect and manage data on the output quantities and qualities of safety interventions implemented to achieve the desired focus on results? Safety engineering treatments? Police operations? Educational activities? Promotional activities? Driver training? Vehicle testing? Emergency medical services? For each category of vehicles and safety equipment (private, commercial, public, helmets) are systematic and regular safety rating surveys undertaken to quality assure adherence to the specified safety standards and rules to achieve the desired focus on results? Vehicle safety rating? Helmet testing? For each category of post-crash service (pre-hospital, hospital, long-term care) are systematic and regular surveys undertaken to quality assure adherence to the specified standards and rules to achieve the desired focus on result? Are systems in place to monitor and evaluate safety performance against targets regularly to achieve the desired focus on results? Do all participating agencies and external partners and stakeholders have open access to all data collected? Source: Bliss, Tony and Jeanne M. Breen. 2009. Country Guidelines for the Conduct of Road Safety Management Capacity Reviews and the Specification of Lead Agency Reforms, Investment Strategies and Safe System Projects. Washington, D.C. : World Bank Group. http://documents.worldbank.org/ curated/en/712181469672173381/GRSF-Country-Implementation-Guidelines. 85 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS Annex B. Interview Question Sets QUESTIONS ON ORGANIZATION AND CAPACITY TO ROAD SAFETY ANALYSIS GROUP/LEAD ROAD SAFETY AGENCY 1. Organization of research/policy-making units: a. To which ministry/governmental department does the unit belong? b. How many staff work at unit and what is their background? c. Are people working full time for the unit? d. For how long do people in general stay in the unit? e. How are staff members trained for their job? f. How do you see the cooperation with different stakeholders? TO THE POLICE 1. Organization of the police: a. Are there dedicated police units for road safety? b. Ratio of police staff dedicated to traffic safety relative to population? c. What is the geographical scope of your competence? All roads? Only urban? QUESTIONS REGARDING REGISTRATION OF CRASH DATA TO POLICE, ROAD SAFETY ANALYSIS GROUP, PLUS OTHER ACTORS (SUCH AS STATISTICS OFFICE) 1. Which definitions do you apply (and why are certain choices made)? a. For road traffic crashes (including single-vehicle crashes, including crashes without motorized vehicles, and including crashes on nonpublic roads) b. For fatalities c. For serious road injuries 2. Notification: a. How are the police notified that a road traffic crash has occurred? (Central emergency number? Who notifies the police?) b. Could there be crashes you are not informed of (crashes with only one vehicle involved, or without motor-vehicle involved? What if someone has no insurance or if the casualty is family and they do not want to be prosecuted, etc.)? 86 MOBILITY AND TRANSPORT CONNECTIVITY SERIES 3. Crash investigation: a. Do the police have the resources to attend every crash scene? (Remoteness, geographic or climate conditions, too many crashes, etc.) b. Which characteristics are reported (form available)? i. About the casualties (severity, person characteristics, seatbelt use, helmet use, alcohol use) ii. About the crash (conflict type, maneuver, location, circumstances, speed, alcohol, etc.) iii. About the vehicles involved (types of vehicles involved, characteristics) iv. Are all participants in a road crash tested for alcohol? v. How are other characteristics determined (in particular severity, causation factors, protective equipment)? 4. Transmission of data: a. How do the police report a crash? (Standard form? Online/on paper? Uniform throughout the country?) b. Is the person who fills out crash registration form the same as the one attending the scene? (If not, which information is the entry based on?) c. Who is responsible for finalizing the crash registration form? Is the completed registration form checked by someone at the local police station? d. If on paper: Who is responsible for entering data in the database (Is registration form sent to central location or is this done at local police station, and by whom?) e. Is someone at the local police station checking whether all registration forms are com- pleted and sent to/imported into the central database? f. Is double counting possible? And if so, what is done to avoid it? g. Do you check with the hospital whether a casualty has died in the hospital and if so, how is information on injury severity updated in database? 5. Are police officers trained in reporting road traffic crashes? If yes: a. What kind of training, how many days? b. By whom? c. All police officers? d. Is there any follow-up? e. Are there refresher courses? 87 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS TO LOCAL POLICE 1. Consequences of not reporting/registering, and own use of data: a. Are you obliged to report all road traffic crashes/all serious/fatal road traffic crashes? By law? By a higher-order institution? b. Does someone outside the police station perform checks on crashes reported by your police station? c. Do you get any feedback on registration of crashes/reported crashes? d. What happens if you do not report (any) crashes? e. What are the reasons for not reporting crashes or for not completing the report in the central database? (in case relevant) f. Do you use data on reported crashes yourself? Or are you aware of data being used by local decision makers? TO CENTRAL INSTITUTION 1. Checking, linking, augmenting: a. Are data linked to other databases (vehicle, hospital, judicial, etc.) to enrich the information? b. Are data linked or compared to other databases (hospital, vital statistics, coroner reports, mortality statistics) to check the numbers? c. Are plausibility checks executed? Which? d. Are data inconsistencies, etc., reported back to those who collected them in the first place? 2. Registration level of fatalities and serious road injuries: a. Are you aware of under-reporting? If there is a difference between reported numbers of fatalities and other estimated numbers (WHO, vital statistics, etc.) discuss this here. b. Are all road users reported? Vulnerable road users? Single vehicle crashes? Even those that might not benefit from insurance? Why are choices made? c. Are all regions covered? All days of the week? Daytime as well as nighttime? If there are regions and/or periods for which the results of data analysis look suspicious, discuss this here. d. Are victims who die later on in hospital included in fatalities? How is this organized? 88 MOBILITY AND TRANSPORT CONNECTIVITY SERIES 3. Do the police have data on preventive interventions? a. Number of checks conducted/hours spent b. Number of fixed/mobile radars c. Number of tickets and fines QUESTIONS ABOUT SURVEILLANCE OF ROAD TRAFFIC INJURIES HOSPITAL STAFF, MINISTRY OF HEALTH, ROAD SAFETY ANALYSIS GROUP, STATISTICS OFFICE, ETC. 1. Practice concerning medical treatment of road traffic casualties: a. Which hospitals or health facilities treat road traffic casualties? b. Are all road traffic casualties treated in hospitals? What if someone is not insured? c. How is it decided to which hospital a road traffic casualty is brought? d. How do road traffic casualties arrive at the hospital? Mainly/all by ambulance? Public or private ambulance services? e. What happens then? First emergency department, then admission to the hospital if needed? 2. Injury surveillance system: a. How do hospitals collect data in the injury surveillance system? b. Are procedures uniform throughout the country? c. How are data from different hospitals combined? d. Is the system exclusively used for the collection of data on road traffic casualties? If not, how can road traffic casualties be identified? e. How is it determined whether a patient is a road traffic casualty? What is the definition of a road traffic casualty? f. Which variables or characteristics are included in the injury surveillance system? g. Do you perform checks? If so, which? (missing records, completeness of records, inconsistencies) h. Do you provide feedback to hospitals concerning the data they provide? (completeness, inconsistencies, use of data?) 3. Registration level: a. Are all hospitals and health centers obliged to report casualties in the injury surveillance system? What happens if they fail to report? b. Which casualties are recorded in the system? People who are admitted in hospital? Or also outpatients? 89 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS c. Are there any benefits for the hospital from reporting road traffic casualties in the injury surveillance system? d. Is the system used for purposes other than the registration of road traffic casualties? e. Do you have an indication of the level of completeness or under-registration? f. Is there a register for all ambulance trips? Would this information be useful for checking the registration of road traffic casualties? TO THE LOCAL HOSPITAL 1. Practical implementation: a. Do you report road traffic casualties that you treat in the injury surveillance system? If not, why not? b. Which casualties? Only casualties that are admitted, or also emergency department? c. Does the surveillance system also include other patients? If so, how do you mark road traffic casualties/how can road traffic casualties be selected? d. What is your definition of a road traffic casualty? e. How does it work in practice? When do you report? Directly, or on the basis of information from another system or registry? f. Who enters the information in the injury surveillance system? Are these people trained in how to do this? How, any follow-up, refresher training? g. Which variables or characteristics are included in the injury surveillance system? h. How is injury severity assessed? 2. Consequences of not or incorrectly reporting: a. Are you obliged to report casualties in the injury surveillance system? What happens if you do not report? b. Does someone check (inside or outside the hospital) if casualties are reported and if reporting is correct? c. Are there any benefits for the hospital from reporting road traffic casualties in the injury surveillance system? d. Is the system used for purposes other than the registration of road traffic casualties? 3. Other information: a. Do you report back to the police when a road traffic casualty has died so it can be included as a fatality in the police record? 90 MOBILITY AND TRANSPORT CONNECTIVITY SERIES QUESTIONS ON OTHER ROAD SAFETY DATA MINISTRIES AND DEPARTMENTS FOR TRANSPORT AND INTERIOR, STATISTICS OFFICE, ROAD SAFETY ANALYSIS GROUPS, AND ADVOCACY GROUPS 1. Information on travel/mobility/vehicle registration: a. What kinds of data are available? (demographics, vehicle registration, vehicle miles trav- eled, mobility survey) b. Check issues listed for each variable that is collected 2. Information on Safety Performance Indicators (SPIs): a. Are you familiar with the concept of SPIs? b. Which SPIs could be relevant for your country? c. Which information on SPIs is already available and how reliable is this information? d. For which SPIs could information be collected and how? QUESTIONS ABOUT ROAD SAFETY DATA STORAGE AND ACCESSIBILITY 1. Crash database: a. What is the architecture of the crash database? b. Who has access? c. Is there a data warehouse that enables combining crash data with exposure data or SPIs? Is the information joined or linked? d. Are there any standard outputs (dashboards, visualizations) produced automatically? QUESTIONS ON ROAD SAFETY DATA USE TO DECISION MAKERS, ROAD SAFETY ADVOCACY GROUPS, ROAD SAFETY ANALYSIS GROUPS, ENGINEERS, AND TRANSPORT PLANNERS 1. Analysis of data: a. Which analyses do you perform? i. Do you cross two or more crash variables to receive specific numbers (for example, children dying as pedestrians in the period before school starts)? ii. Do you monitor the development of casualties within particular groups (for example, motorcyclists, young adults, etc.)? iii. Do you produce maps with crashes? iv. Do you relate crashes to particular road characteristics? v. Do you evaluate countermeasures (such as in pre- and post-studies)? 91 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS b. How do you determine which questions to investigate? c. Do you have standard outputs that inform other stakeholders? 2. Road safety policy making: a. How is road safety policy making organized? Who is responsible for: i. Infrastructural measures ii. Regulation (in relation to road user behavior and in relation to vehicles) iii. Enforcement b. Do these authorities base their measures on road safety data? i. Which information? ii. Who provides it? iii. How do they use it? c. What measures are taken to assess under-reporting of fatalities and serious injuries? If applicable, what is done about it? d. Which kinds of data or analysis would they like to see? EVALUATION TO ALL STAKEHOLDERS 1. Do you have any suggestions for improving registration of fatalities and serious injuries? 2. Which other road safety data would you consider most important to collect? 3. How could the use of road safety data be improved? 92 MOBILITY AND TRANSPORT CONNECTIVITY SERIES Annex C. Methodological Issues for Collecting Travel and SPI Data MOBILITY DATA • Response rate or selective responses (in particular A good overview of exposure data for road safety for postal surveys) analyses with practical examples for their collection epresentativeness of time: seasons, week ver- • R can be found in the SafetyNet project as discussed in sus weekend, vacation, among other time-related Yannis et al. (2005). factors TRAVEL DISTANCES ata quality: possible inaccuracies in the estima- • D The gold standard for relating crashes and casual- tion of distances ties to the calculation of risks is the distance traveled (either per vehicle or per person). Typically, participants fill in a travel diary for one day, for which they give detailed reports (time, purpose, • Desirable variables for travel distances: road user travel mode, distance, and often also start and end group and vehicle type, and road type. points). The collection should be spread so that the four seasons are all represented, every day of the A number of methods can be used to estimate travel week, all hours of the day. The estimation of distances distances: and travel times by the travelers themselves is very unreliable. As a solution, rather than sampling per- Surveys sons willing to fill in a questionnaire, one could sam- A representative sample of the population is required ple persons willing to install an application on their to fill in a mobility diary—usually for one day. Partici- smartphone to measure distances, routes, and travel pants indicate departure and arrival time for each trip speeds. As this is a rather new technique, not much made, along with the origin and destination. Mobility experience exists in countries with a long tradition of surveys are, for the moment, the only way to esti- recording travel behavior. Likely issues include rep- mate the distances traveled by vulnerable road users. resentativeness (how to record trips undertaken by Usually, the surveys offer all types of disaggregation persons who do not use smart phones), reliable iden- in terms of person characteristics, including age, gen- tification of the travel mode, and privacy concerns. der, vehicle type, trip purpose, and more. For more information see, “Innovation of the Dutch When conducted online, surveys have become rela- National Travel Survey: Implementation of the New tively cheap, though representativeness can become Design” (Smit, Mol, and van der Waard 2017) pre- an issue. Possible issues include the following: sented at the European Transport Conference in Bar- celona, Spain. • Representativeness of the sample: o Rural areas (for all kinds of surveys) Travel counts o Elderly people (for online surveys) Traffic count systems operating in most countries o Working people (for phone surveys or allow for continuous measurement of traffic volumes door-to-door interviews) over time. The measurement sites could be more or 93 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS less representative of the entire road network exam- express risk (such as casualties per ton of fuel). While ined. Two main approaches are used to derive vehi- this can show real progress in countries with rapidly cle kilometers from counts: one based on weighted growing motorization, it is inadequate in countries counts, in which a site is assumed to be represen- with minor changes in the distances driven as the tative of a number of other sites, and the other trend is also influenced by changes in fuel efficiency. approach, which uses statistical models to estimate Many countries use this method, but mostly in com- the counts for the nonmeasured sites. In both cases, bination with other methods, including Germany and the estimated counts are multiplied by the length of France. Data on fuel efficiency in France can be found the sections to obtain aggregate vehicle kilometers. in Ricrorch and Sarron (2018, 168–75). • Counting devices: pneumatic tubes, cameras, • Possible variables: fuel type (diesel versus RADAR, LIDAR gasoline) • Desirable variables: road type, area type, region, • Issues: changes to the energetic efficiency of vehicle type (if counting device can differentiate) vehicles • Issues: representativeness of measurement loca- tion; estimating counts for nonmeasured locations OTHER TRAFFIC INDICATORS Odometer readings Road length The information on vehicle distances traveled can • Desirable variables: road type, area type, region be based upon the vehicles’ technical inspections, if mandatory at a particular interval (for example, • Issues: often unavailable for local roads. More- annually). When a vehicle is inspected, the distance over, it is difficult to establish a uniform interna- traveled is registered and compared with the reading tional classification of the types of roads outside from the previous inspection. Knowing the type of urban areas that goes beyond indicating if it is a vehicle and the total numbers of that type, it is possi- motorway or not. ble to give an estimate of distance traveled by type of vehicle. Vehicle fleet (from vehicle register) • Desirable disaggregation: vehicle type, vehicle age • Desirable disaggregation: vehicle type; vehicle age (engine size) • Issues: Foreign vehicles are not included; no infor- mation when or where kilometers are driven • Possible issues: o No inclusion of foreign vehicles Fuel consumption o The inclusion of new vehicles (this is Driven kilometers can be estimated on the basis of typically good if the purpose of the data- fuel consumption (see SafetyNet D2.1 in Yannis et base—for example, taxation, insurance, al. 2005). However, the fuel efficiency of different technical inspection—is mandatory for types of vehicles and their share in the fleet must be all vehicles known. Raw fuel consumption can also be used to 94 MOBILITY AND TRANSPORT CONNECTIVITY SERIES o Are all types of vehicles included? What for alcohol use is probably the Driving under the about motorcycles and mopeds? Influence of Drugs, Alcohol and Medicines (DRUID) o Are identification codes unique? Can project (see Houwing et al. 2011). there be duplicates? (What about lost license plates?) ROADSIDE SURVEYS o Are scrapped vehicles removed? Because many SPIs require roadside observation, some basic rules should be checked when evaluating Driver population (from license registration) the data collection for an indicator. Ideally, a sampling plan to address time and place of measurement will • Desirable disaggregation: age, gender, vehicle be repeated regularly, say, annually, in the same way. type (nationality, experience) • Possible issues: Time: Check for a reason to believe that rates differ between night and day and week versus weekend o Foreigners are not included (such as for alcohol) or seasons (such as pedestrian o Deceased drivers (or withdrawals): Are or bicycle counts). If so, significant periods must be they removed from the register? represented according to their actual share of traffic. o Could there be duplicate entries (such as If no reason exists to believe the rates differ (distrac- for upgraded licenses)? tion and seatbelt use, for example), daytime mea- o Hierarchies—for example, car licenses surements with one observation period per year are can also permit the riding of mopeds acceptable. or motorcycles, making it impossible to estimate the number of drivers for each Place: Distinguish different types of cities (such as category. capital, large city, or town) as well as between rural roads and motorways. These five main categories of public road network should be included more or less SAFETY PERFORMANCE INDICATORS proportionally to their share of traffic. However, at Although relatively old, the methodology presented least 1,000 vehicles should be observed per category. by the EC project SafetyNet (Hakkert, Gitelman, and Weighting: Often the sampling plan includes the Vis 2007) can still be considered the international same number of vehicles per relevant level, such as state-of-the-art for many of the safety performance road types, period types, or regions, rather than dif- indicators (SPIs), including those for speed, occupant ferent sample sizes according to their share in traffic. protection, and protective gear. In this case each level has to be weighted according to its actual size when calculating the national aver- For speed data, this is however rather outdated, age. Sometimes, certain types of road sites, periods, because nowadays speed data are available from or vehicle types are oversampled on purpose. For providers of navigation services. They are even pro- example, for driving under the influence of alcohol vided for free by big data providers such as Google or and drugs, weekend nights are considered particu- Uber. These companies provide the mean speed per larly relevant. So, although only a small proportion of road segment, which can be matched to crash data. traffic takes place in these periods, the sample for this The best reference for conducting road-site surveys period should be large enough to allow for analysis of 95 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS driver characteristics. Again, in the overall analyses, measured should be completed before switching to such as for different periods throughout the week, the next vehicle. Skipping vehicles is acceptable, but weighting has to be applied to correct for oversam- incomplete records should be avoided. pling. A good description of necessary sample sizes, measurement errors and weighting can be found in Regular measurement of the key performance indica- chapter 2 of Hakkert, Gitelman, and Vis (2007). tors is essential for monitoring road safety. To make this comparable over time, the same methodology Observations: Vehicles should be selected randomly should be maintained. For this reason, the first mea- for observation. This point is particularly important surement should be carefully planned because later to stress if the measurement is conducted in coop- improvements to the methodology will always affect eration with the police, because it runs counter to the comparability of successive measurements. their usual practice. Police would usually focus on suspicious-looking drivers, either on the basis of their Next to regular monitoring, countermeasures should behavior or because of a dominant offender profile, be evaluated by conducting measurements of the which does not result in a representative measure- related SPIs before and after implementation. If ment. Rather, vehicles should be checked strictly in regular measurements are in place, at least one of the order they arrive at the measurement location. these pre- and post-measurements can be conducted For each vehicle selected, reporting of all variables within the regular monitoring measurement. References Hakkert, A. S., V. Gitelman, and M. A. Vis, eds. 2007. “Road Safety Performance Indicators: Theory. Deliverable D3.6 of the EU FP6 Project Safetynet.” Report, Loughborough University, United Kingdom. https://hdl.handle.net/2134/4952. Houwing, Sjoerd, Marjan Hagenzieker, René Mathijssen, Inger Marie Bernhoft, Tove Hels, Kira Janstrup, Trudy Van der Lin- den, Sara-Ann Legrand, and Alain Verstraete. 2011. “Prevalence of Alcohol and Other Psychoactive Substances in Driv- ers in General Traffic, Part II: Country Reports.” DRUID (Driving under the Influence of Drugs, Alcohol and Medicines). http://hdl.handle.net/1854/LU-1988588. Ricrorch, Layla and Clotilde Sarron. 2018. “Les comptes des transports en 2017 : 55e rapport de la Commission des comptes des transports de la Nation.” Le service de la donnée et des études statistiques (SDES), La Défense CEDEX, France. https://www.statistiques.developpement-durable.gouv.fr/sites/default/files/2018-11/datalab-42-rapport-compt- es-transports-2017-aout2018.pdf. Smit, R., M. Mol, and J. van der Waard. 2017. “Innovation of the Dutch National Travel Survey: Implementation of the New Design.” Paper presented at the European Transport Conference, Barcelona, Spain, October. https://www.kimnet.nl/ publicaties/papers/2017/10/04/innovation-of-the-dutch-national-travel-survey-implementation-of-the-new-design. Yannis, George, E. Papadimitriou, P. Lejeune, V. Treny, S. Hemdorff, R. Bergel, M. Haddak, P. Holló, J. Cardoso, F. Bijleveld, S. Houwing, T. Bjørnskau. 2005. “State of the Art of Risk and Exposure Data.” Deliverable 2.1 of the EC FP6 project SafetyNet, European Commission, Brussels, Belgium. https://www.swov.nl/en/publication/ state-art-report-risk-and-exposure-data. 96 MOBILITY AND TRANSPORT CONNECTIVITY SERIES Annex D. Examples of Database Structure The structures of crash databases can vary and often is integrated in a powerful information technology result from their history and from the different play- (IT) system; diagram C describes its dataflow. ers and organizations led to build and then develop them over time. In addition, databases often retain The Data for Road Incident Visualization Evaluation in their structure the memory of old technical con- and Reporting (DRIVER) is largely based on data ele- straints, such as the capacity to store, transmit, or ments normally collected by the police, but can be address data. customized to reflect a particular country’s context and needs. DRIVER allows for integration of crash Most crash databases are structured according to the data with other types of data, such as health data principles of diagram A shown in figure D.1. How- from injury surveillance systems, vehicle registration ever, according to the Common Accident Data Set data, driver licensing data, and road infrastructure (CADaS), diagram B illustrates the European Union data. Because of this flexibility, the vehicle, person, Care Database structure. The French database TRAxy and environment details can be linked to the crash. Figure D.1. Crash Database Structures Diagram A Environment Crash Crash Crash Environment Vehicles Environment Persons Persons Vehicles Vehicles Involved Involved Persons Involved Source: Original graph produced for this publication. 97 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS Diagram B Accident Traffic Unit “ VARIABLE_NAME REF_LINK VARIABLE_NAME REF_LINK ACCIDENT_ID A-1 ACCIDENT_ID A-1 1 1 1 COUNTRY_CODE E_ROAD R-3 YEAR FUNC_CLASS_1ST_ROAD R-5 ACCIDENT_DATE A-2 FUNC_CLASS_2ND_ROAD R-6 ACCIDENT_TIME A-3 ANNUAL_DAILY_TRAFFIC_1ST_ROAD R-7 NUTS A-4 ANNUAL_DAILY_TRAFFIC_2ND_ROAD R-8 LAU A-5 SPEED_LIMIT_1ST_ROAD R-9 WEATHER_CONDITION A-6 SPEED_LIMIT_2ND_ROAD R-10 LIGHT_CONDITIONS A-7 MOTORWAY R-11 ACCIDENT_WITH_PEDESTRIANS A-8 ROAD_SURFACE_CONDITIONS R-16 ACCIDENT_WITH_PARKED_VEHICLES A-9 ROAD_OBSTACLES R-17 SINGLE_VEHICLE_ACCIDENTS A-10 CARRIAGEWAY_TYPE R-18 AT_LEAST_2_VEH_NO_TURNING A-11 NUMBER_OF_LANES R-19 AT_LEAST_2_VEH_TURN_CROSS A-12 EMERGENCY_LANE R-20 LATITUDE ‘ R-1 ROAD_MARKINGS R-21 LONGITUDE ‘ R-2 BRIDGE R-22 E_ROAD_KILOMETRE ‘ R-4 WORK_ZONE_RELATED R-23 URBAN_AREA ‘ R-12 REGISTRATION_COUNTRY R-24 JUNCTION ‘ R-13 ROAD_CURVE R-25 REL_TO_THE_AT_GRADE_JUNC ‘ R-14 ROAD_SEGMENT_GRADE R-26 JUNCTION_CONTROL ‘ R-15 Traffic Unit “ Accident VARIABLE_NAME REF_LINK VARIABLE_NAME REF_LINK ACCIDENT_ID A-1 ACCIDENT_ID A-1 N TRAFFIC_UNIT_ID U-1 TRAFFIC_UNIT_ID U-1 1 N PERSON_ID P-1 TRAFFIC_UNIT_TYPE U-2 VEHICLE_SPECIAL_FUNCTION U-3 YEAR_OF_BIRTH P-2 TRAILER U-4 GENDER P-3 ENGINE_POWER U-5 NATIONALITY P-4 ACTIVE_SAFETY_EQUIPMENT U-6 INJURY_TYPE P-5 VEHICLE_DRIVE U-7 ROAD_USER_TYPE P-6 MAKE U-8 ALCOTEST P-7 MODEL U-9 ALCOTEST_SAMPLE_TYPE P-8 REGISTRATION_YEAR U-10 ALCOTEST_RESULT P-9 TRAFFIC_UNIT_MANOEUVRE U-11 ALCOTEST_LEVEL P-10 FIRST_POINT_OF_IMPACT U-12 DRUG_TEST P-11 FIRST_OBJECT_HIT_IN U-13 DRIVING_LICENCE_ISSUE_DATE P-12 FIRST_OBJECT_HIT_OFF U-14 DRIVING_LICENCE_VALIDITY P-13 INSURANCE U-15 SAFETY_EQUIPMENT P-14 HIT_AND_RUN U-16 POSITION_IN_ON_VEHICLE P-15 REGISTRATION_COUNTRY U-17 PARTIC_DISTR_BY_DEVICE P-16 PSYCHO_PHYS_IMPERMENT P-17 ‘ : Fields moved from Road teble to Accident table TRIP_JOURNEY_PURPOSE P-18 “ : Traffic Unit can also be a Vehicle or a Pedestrian Source: European Commission 98 MOBILITY AND TRANSPORT CONNECTIVITY SERIES Diagram C Source: French National Interministerial Road Safety Observatory. 99 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS Annex E. Sample Terms of Reference 1. OBJECTIVE 3.2. Consult each ministry related to crash data, The primary objective of the proposed consulting SPIs, and mobility data reporting and analysis, services is to review and assess crash data and road and collect feedback regarding current proce- safety data collection and analysis and develop a road dures and systems. map for improving data in the country. 3.3. Evaluate the institutional arrangement for reporting, recording, analyzing, and shar- 2. SCOPE OF SERVICES ing data and the extent to which the current The Consultant shall review the prevailing system of arrangement meets agency requirements for road crash data and road safety data collection and analysis and understanding of road safety analysis, including conducting research and under- problems. taking a literature review of published studies and 3.4. Identify and review relevant institutional and reports, reviewing existing documentation, forms, legal policy instruments related to road crash and reports, assessing legal and policy instruments, data collection, analysis, and sharing. inspecting crash data, assessing safety performance indicators (SPIs) and mobility data, conducting con- 3.5. Assess current technical resources and staff sultations and interviews with relevant stakeholders, capacities of each ministry in relation to crash and drafting a report enumerating and discussing data collection. findings and analysis as well as recommending the 3.6. Review the crash data forms used by the differ- way forward in improving data. ent ministries and identify and compare data elements and definitions used and methods of 3. MAIN TASKS collection. The Consultant shall conduct a detailed review of crash data and road safety data in the country. This 3.7. Review earlier and ongoing initiatives and includes systems used by the different ministries reviews on data, assess their findings, and such as the police, the health sector, civil registry, identify lessons learned and challenges. road transport, among other relevant offices. 3.8. Examine actual on-the-ground reporting practice and identify strengths, gaps, and 3.1. Conduct a preparatory research and review, challenges. which includes identifying stakeholders and 3.9. Identify gaps in current arrangement, espe- government structure, reviewing policy and cially with respect to underreporting of crashes, legal instruments, assessing existing documen- fatalities and serious injuries, compliance with tation, forms, and reports, inspecting data, and reporting formats and procedures and rec- assessing SPIs, and mobility data. ommend improvements to institutional and reporting arrangements and procedures. 100 MOBILITY AND TRANSPORT CONNECTIVITY SERIES 3.10. Recommend necessary modifications to cur- 4. REPORTING REQUIREMENTS rent and planned procedures and systems for recording (including reporting process and 4.1. The Consultant is expected to carry out the report form), analyzing, and reporting road assignment tasks as stipulated in the terms of crashes, and recommend methodologies for reference (ToR), in very close coordination with the efficient and accurate entry of data. concerned government agencies. 3.11. Based on the assessment, prepare a report 4.2. The Consultant shall complete the outputs and summarizing findings and analysis as well as deliverables based on the schedule displayed in identifying recommendations for improving the table below. crash data, SPIs, and mobility data collection 4.3. The duration of the services will be two-and-a- and analysis. half (2.5) months from the time of the contract signature. No. Required deliverable and/or output Due timing (from mobilization) 1 Inception report (IR), inclusive of all tasks with detailed work program 15 days 2 Data review draft report 2 months 3 Data review final report 2.5 months 5. TEAM COMPOSITION AND QUALIFICATIONS The professional qualifications, skills, and experience required are as follows: No. Position Minimum qualifications Specific required expertise 1 Team leader Graduate qualifications in Sound in-depth knowledge of national or international engineering, law, economics, findings and directions in modern road safety management administration, management, principles and the “safe system” approach to road safety. or equivalent/relevant field. Should have about 10 years of experience conducting sci- Specialist high-level qualifica- entific analyses of road environment, vehicle and human tions relevant to Road Safety factors contributing to road crashes and injuries. management and coordination Experience of being a team leader on similar assignment is functions is preferable. desirable. 101 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS No. Position Minimum qualifications Specific required expertise 2 Deputy team Graduate qualifications in civil, About eight years of experience in data information and leader cum IT mechanical, or transport engi- management systems including user friendly interfaces; specialist neering, or computer science or extensive experience in managing complex information equivalent. technology (IT) projects, across a range of public sector agencies and levels of administration, at least two of which should be in developing and transitional countries. Experience with crash analysis systems or road safety infor- mation management systems is desirable. 3 Road safety Graduate qualifications in engi- Experience in highway/general policing/enforcement with specialist neering, science, economics, minimum overall professional experience of eight years. administration, management, Shall be highly experienced in leading implementation of or equivalent/relevant field. large road safety programs. 102 MOBILITY AND TRANSPORT CONNECTIVITY SERIES PHOTO CREDITS Cover page and page 6: By metamorworks, Shutterstock Page 8: By Tonktiti, Shutterstock Page 11: By metamorworks, Shutterstock Page 16: By Photo Spirit, Shutterstock Page 23: By Milkovasa, Shutterstock Page 34: By Trevor Samson, World Bank Page 38: By Reshetnikov_art, Shutterstock Page 39: By Virrage Images, Shutterstock Page 43: By Pair Srinrat, Shutterstock Page 47: By Sarah Farhat, World Bank Page 55: By Rob Hyrons, Shutterstock Page 61: By smartman, Shutterstock Page 63: By Matej Kastelic, Shutterstock Page 64: By Grand Warszawski, Shutterstock Page 66: By Jaromir Chalabala, Shutterstock Page 67: By Simone D. McCourtie, World Bank Page 70: By Trzykropy, Shutterstock Page 77: By Gorodenkoff, Shutterstock Page 78: By Photographee.eu, Shutterstock Page 82: By Tauno Tõhk, 2.0 Generic (CC BY 2.0) 103 GUIDELINES FOR CONDUCTING ROAD SAFETY DATA REVIEWS MOBILITY AND TRANSPORT CONNECTIVITY SERIES: 2021 REPORTS Accelerating Digitalization: Critical Do Speed Limit Reductions Help Road Actions to Strengthen the Resilience Safety?: Lessons from the Republic of of the Maritime Supply Chain Korea’s Recent Move to Lower Speed https://openknowledge.worldbank.org/ Limit on Urban Roads. handle/10986/35063 https://openknowledge.worldbank.org/ Available also in French. handle/10986/36109 World Bank. 2021. Mitra, Sudeshna; Job, Soames; Han, Sangjin; Eom, Kijong. 2021. Closing the Gap: Gender, Transport, Electrification of Public Transport: A and Employment in Mumbai Case Study of the Shenzhen Bus Group. https://openknowledge.worldbank.org/ https://openknowledge.worldbank.org/ handle/10986/35297 handle/10986/35935 World Bank. 2021. World Bank. 2021. Connectivity for Human Capital: To Pave or Not to Pave: Developing a Realizing the Right to Education and Framework for Systematic Decision- Healthcare through Improved Public Making in the Choice of Paving Transport in African Cities Technologies for Rural Roads https://openknowledge.worldbank.org/ https://openknowledge.worldbank.org/ handle/10986/35185 handle/10986/35163 World Bank. 2021. World Bank. 2021.