|Year : 2013 | Volume
| Issue : 1 | Page : 25-30
Variations in sub-national road traffic fatality trends in a low-income country
Junaid A Bhatti1, Ajmal Khan Khoso2, Hunniya Waseem3, Uzma Rahim Khan4, Junaid A Razzak4
1 Section of Injury and Violence Prevention, Public Health Solutions Pakistan, Lahore, Pakistan; Addiction Research Program, Douglas Hospital Research Centre, Montreal, Canada
2 Office of N-5, Sector I South Zone, National Highway and Motorway Police, Karachi, Pakistan
3 Section of Injury and Violence Prevention, Public Health Solutions Pakistan, Lahore, Pakistan
4 Department of Emergency Medicine, The Aga Khan University, Karachi, Pakistan
|Date of Web Publication||22-Mar-2013|
Junaid A Bhatti
Public Health Solutions Pakistan, 269-Y Block, Office 1, First Floor, DHA Phase 1, Lahore, Pakistan
Source of Support: None, Conflict of Interest: None
| Abstract|| |
Background: In most low- and middle-income countries (LMICs), road traffic fatality (RTF) trends are presented in aggregated form at the national level. This practice omits important information regarding RTF risk at sub-national levels.
Objective: This ecological study assesses the extent of RTF variations at different sub-national levels in Pakistan, a low-income country.
Materials and Methods: Based on official statistics, significant variations in three RTF indicators i.e. per population, per registered vehicles, and per crash were compared by regression analyses at two sub-national levels i.e. provincially (2000-2009) and district-wise (2004).
Results: The national RTF counts are based on data from four provinces. From 2000 to 2009, RTF per population and per registered vehicles decreased in all provinces except Balochistan. RTF per crash in Punjab decreased from 0.61 to 0.56 (beta coefficient (β) year = -0.0082, P = <0.001), whereas in Balochistan it increased from 0.40 to 0.58 (β year = 0.0708, P = <0.001) over the same period. District-level comparisons were possible only in Punjab where RTF per crash varied from 0.25 to 2.15 and correlated (β = 0.50, P = 0.003) with RTF per population.
Conclusions: Sub-national RTF surveillance is necessary in LMICs like Pakistan in order to prioritize available resources on high-risk jurisdictions such as the Balochistan province and districts of Punjab where high RTF per population and per crash exist.
Keywords: Balochistan, Punjab, Sindh, Traffic accidents
|How to cite this article:|
Bhatti JA, Khoso AK, Waseem H, Khan UR, Razzak JA. Variations in sub-national road traffic fatality trends in a low-income country. Int J Crit Illn Inj Sci 2013;3:25-30
|How to cite this URL:|
Bhatti JA, Khoso AK, Waseem H, Khan UR, Razzak JA. Variations in sub-national road traffic fatality trends in a low-income country. Int J Crit Illn Inj Sci [serial online] 2013 [cited 2020 Jul 3];3:25-30. Available from: http://www.ijciis.org/text.asp?2013/3/1/25/109415
| Introduction|| |
Road traffic crashes (RTCs) are a largely ignored public health problem in low- and middle-income countries (LMICs),  despite the fact that they account for more than 90% of the road traffic fatalities (RTF) worldwide.  It was estimated that between 2000 and 2020, RTF would increase by 80% in LMICs and decrease by 28% in high-income countries (HICs).  The United Nations have, therefore, proclaimed the Decade of Action for Road Safety 2011 to 2020 to first stabilize and then decrease the projected RTF worldwide, particularly in LMICs. 
RTF indicators e.g. per population, per registered vehicles, per crash etc. are important surveillance tools to monitor road safety at national and sub-national levels. , Previous work using international data collected by the World Health Organization and International Road Federation showed that variations in RTF indicators could be explained by macroeconomic factor such as gross national income (GNI) as well as safety-specific interventions such as quality of road infrastructure, use of safety equipment (e.g. seatbelts etc.), renewal of car pool, driver education, enforcement of standard laws and access to quality emergency care. , Some studies have reproduced these results at sub-national level as well, suggesting that comparisons of different RTF trends could be useful in prioritizing safety interventions between jurisdictions. ,
In most LMICs, RTF trends are presented as aggregated statistics, usually as yearly RTF counts while sub-national variations in RTF trends are rarely analyzed to target preventive interventions. , This was evident in recent global status report where many LMICs did not present their RTF data by sex, vehicle type, or other important variables including sub-national distributions, all of which are used in HICs to set priorities for interventions.  The reason for publishing aggregate RTF data in LMICs could be the difficulties inherent to data collection practices in cases where multiple agencies are involved.  For instance, RTF data is collected mostly by a crime-focused police, whereas exposure data such as population, registered vehicles, and kilometers-travelled is collected by transport-related agencies. , It is likely that the jurisdictional establishment of involved agencies might be different at sub-national levels, and therefore result in the publication of aggregated RTF data only. ,, Furthermore, most LMICs continue to lack specialized lead road safety agencies, an additional barrier in comparing RTF trends by user characteristics and levels including sub-national levels.  As few studies explored sub-national comparisons of RTF in LMICs, this ecological study is therefore aimed to highlight the availability and variations of sub-national RTF trends in Pakistan, a low-income country.
| Materials and Methods|| |
Setting and design
Pakistan has an estimated population of over 177 million (Jan 1, 2011).  Official estimates from 2000 to 2009 indicate that approximately 7 000 RTF occur each year in Pakistan.  The recent global status report, however, estimated that the actual RTF in Pakistan could be between 28 379 and 76 695 deaths per year.  None of the previous international and national reports had compared the sub-national variations of RTF indicators in Pakistan. ,,
Sub-national levels in Pakistan
Administratively, Pakistan is divided into four provinces and four territories.  The four provinces of Pakistan are Punjab, Sindh, Balochistan, and Khyber Pakhtunkhwa (KP) and the four territories are Azad Kashmir, Gilgit-Baltistan (GB, formerly Northern Areas), the Federal capital (Islamabad), and Federally Administered Tribal Agencies (FATA). ,,,,, Each province and territory has separate yet similarly structured police service as well as vehicle registration authorities with the exception of FATA, which is regulated under a different legislation. , Furthermore, each province is sub-divided into districts and these subdivisions exist at the level of police and vehicle registration authorities as well. , Due to the administrative structure of the provinces and territories of Pakistan, sub-national RTF trends were compared at two levels i.e. provincially and district-wise.
RTF definitions and reporting
In Pakistan, RTC data is principally collected by the local police authorities. , After a RTC, the police or the involved parties register an initial police report explaining the crash circumstances and number of involved road users as well as their injury outcome. , In case of a RTF, the case is registered under the section 320 of the Pakistan Penal Code.  Reports are usually written in the hours following the RTC and patients with severe injuries are not followed for 30 days (as per RTF definition 2 ) unless the criminal case is pursued. The absence of governmental compensations through general insurance and out-of-court settlements may reduce chances of having RTF registered in official statistics.  RTF are recorded on the daily activity registers in each police station as well, and these figures are often communicated to the regional departments (i.e. district or provincial police departments) on a monthly basis. ,, At the end of the year, overall RTC statistics including RTF counts are sent to provincial and federal institutions responsible for collecting and publishing crime statistics.
We used national and provincial statistical reports ,,,,, to extract the data on RTC, RTF, and registered vehicles. The national-level report is published on yearly basis, whereas provinces publish these reports at variable intervals. Copies of these reports were obtained from central repository at the Federal Bureau of Statistics, Islamabad in September 2009. The data was available for two sub-national levels i.e. provincially and district-wise.
The extracted RTF and vehicle count data was presented as a percentage difference i.e. ((counts in 2009 - counts in 2000)/counts in 2000 × 100). National, provincial, and district-wise RTF per population were estimated from the average population growth rates (GR) as available in the census of 1998. RTF per registered vehicles and per crash were computed provincially- and district-wise where available. The provincial RTF indicator trends from 2000 to 2009 were analyzed using Poisson regression models where RTF counts were dependent variables, years were independent variables, and yearly population, vehicle or crash counts were offset variables. Further, linear regression models were used for each province to assess whether RTF per registered vehicles and per crash (independent variables) changed significantly given RTF per population (dependent variables) between 2000 and 2009.
The district-wise analyses differed from above as data was only available from the province of Punjab and for 1 year only (see data availability below). Districts are instead presented as four groups illustrating whether they had lower RTF per population and per crash than provincial RTF per population (i.e. 3.5) and per crash (0.60) for the same year. A linear regression model assessed whether RTF per crash (independent variables) changed significantly given RTF per population (dependent variables) using district-based data where available. SAS version 9.1.3 was used for data handling and analyses.
| Result|| |
RTF data availability
[Figure 1] showed the availability of RTF data in Pakistan. Provincial RTF data was available from 2000 to 2009. The district-wise RTF data was available in Punjab for the year 2004 only. In Sindh data from some of the districts was regrouped into divisions (an intermediate level between province and district) as per police administration structure. No disaggregated data was available for KP and Balochistan. The official statistics did not include the RTF data from the Gilgit-Baltistan, Azad Kashmir, and FATA. The data from Islamabad was aggregated with that of Punjab.
|Figure 1: Road traffic fatality per crash at different sub-national levels in Pakistan (2004)|
Click here to view
Provincial population and vehicle counts trends
In 1998, the population of Pakistan was 132.3 million (Growth Rate [GR] = 2.7%), the province of Punjab including Islamabad 74.4 million (GR = 2.6%), KP 17.7 million (GR=2.8%), FATA 3.2 million (GR=2.2%), Sindh 30.4 million (GR=2.8), and Balochistan 6.6 million (GR=2.5%). Punjab (358 inhabitants per sq. km), Sindh (216 inhabitants per sq. km), and KP (166 inhabitants per sq. km) were densely populated compared to the Balochistan province (19 inhabitants per sq. km).  In Pakistan the registered vehicles increased by 38.3% from 4.7 million in 2000 to 6.5 million in 2009. Increases in registered vehicles in provinces from 2000 to 2009 were 35.7% in Punjab (from 2.8 to 3.8 millions) and 41.1% in Sindh (from 1.4 to 1.9 million), 50.0% in KP (from 0.4 to 0.6 million), and 100.0% in Balochistan (0.1 to 0.2 million). District-wise registered vehicles were not available for any of the provinces.
Provincial RTF count trends
From 2000 to 2009, RTF counts decreased by 4.4% (5 130 to 4 907) in Pakistan, by 1.4% in Punjab (2 954 to 2 912), and by 27.7% in Sindh (1 288 to 931) but increased by 6.5% in KP (738 to 786) and by 65.3% (150 to 248) in Balochistan. From 2000 to 2009, RTC counts decreased by 0.04% (9 735 to 9 696) in Pakistan, by 1.4% in KP (2 448 to 2 392), and by 31.4% in Sindh (2 088 to 1 433) but increased by 8.8% in Punjab (4 826 to 5240) and by 122.6% (373 to 431) in Balochistan.
Provincial RTF per population trends
[Table 1] presents RTF trends in Pakistan as well as in the four provinces. RTF per 100 000 inhabitants in Pakistan decreased from 3.7 in 2000 to 3.0 in 2009 (beta coefficient (β) = - 0.0199, P = <0.001). From 2000 to 2009, RTF per 100 000 inhabitants significantly decreased from 3.9 to 3.0 in Punjab (β = - 0.0289, P = <0.001) and from 4.1 to 2.4 in Sindh (β = - 0.0554, P = <0.001) but increased from 2.2 to 3.0 in Balochistan (β = 0.0759, P = <0.001).
|Table 1: Road traffic fatality (RTF) indicators for provinces in Pakistan from 2000 to 2009|
Click here to view
Provincial RTF per registered vehicles trends
RTF per 10 000 registered vehicles in Pakistan decreased from 11.3 in 2000 to 9.1 in 2009 (β = - 0.0188, P = <0.001). From 2000 to 2009, RTF per 10 000 registered decreased from 10.7 to 7.6 in Punjab (β = - 0.0391, P0 = <0.001), from 9.4 to 4.9 in Sindh (β = - 0.0644, P = <0.001), and from 18.1 to 14.2 in KP (β = - 0.0129, P = 0.001) but increased from 9.6 to 11.5 in Balochistan (β = 0.0644, P = <0.001).
Provincial RTF per crash trends
RTF per crash in Pakistan decreased from 0.53 in 2000 to 0.51 in 2009 (β = - 0.0006, P = <0.001). From 2000 to 2009 RTF per crash decreased from 0.61 to 0.56 in Punjab (β = - 0.0082, P = <0.001), increased from 0.62 to 0.67 in Sindh (β = 0.0144, P = <0.001), from 0.30 to 0.33 in KP (β = 0.0124, P = 0.001), and from 0.40 to 0.58 in Balochistan (β = - 0.0708, P = <0.001).
Variations in provincial RTF trends
RTF per population and registered vehicles correlated significantly (0.76 ≥ β ≤ 1.02, P = <0.001) with each other at national and provincial levels [Table 2]. Variations in RTF fatalities per population were not associated with RTF per crash at the national level in Punjab or Sindh, whereas they were significantly associated with RTF per crash in KP (β = 0.19, P = 0.005) and Balochistan (β = 0.26, P = 0.02).
|Table 2: Regressions of road traffic fatality (RTF) per population (Y) on RTF per vehicle (X1) and RTF per crash (X2) at provincial and district levels in Pakistan (2000-2009)|
Click here to view
District-wise RTF trends
Fourteen districts of Punjab had higher than average RTF per population and per crash in 2004 [Table 3]. The highest value of 2.15 RTF per crash was observed for district Mianwali. The linear regression analyses showed that RTF per population were significantly associated with RTF per crash in different districts of Punjab in 2004 (β = 0.50, P = 0.003).
|Table 3: Road traffic fatality (RTF) indicators for districts in the province of Punjab - 2004|
Click here to view
| Discussion|| |
This study highlighted the capacity of RTF data from Pakistan in monitoring different aspects of RTC problem in Pakistan. It showed that three major jurisdictions (FATA, GB, and Azad Kashmir) are not covered in national RTF statistics. Of the monitored jurisdictions, RTF per population decreased in all jurisdictions except Balochistan, where it increased alarmingly over the last decade. RTF per crash only decreased in Punjab, whereas it increased slightly in Sindh and KP and augmented by half in Balochistan. District-wise data was not systematically available. In cases where it was available, such as in Punjab, data indicates high RTF risk areas.
Overall, RTF per crash in Pakistan was twice as high than the neighboring countries like India (0.21) and China (0.22) and about tens of times higher than United States (US) (0.02), France (0.06), and Britain (0.01).  The preferential reporting of severe and fatal RTC could explain relatively high values of RTF per crash in Pakistan.  A secondary analysis  of global status report's data supported the above hypotheses by indicating that the official statistics in Pakistan accounted for about 16.7% of RTF, this proportion was relatively low compared to regional countries such as Iran (89.9%) and Egypt (50.8%). Detailed comparisons of reporting practices in different countries, however, are needed to fully explain relatively high RTF per crash in Pakistan.
The paradoxical evolution of RTF per crash at provincial and district levels in Pakistan was an interesting finding from the perspective of public health practice i.e. the observed differences could be explained by geographical distribution of safety-related factors such as police enforcement, road quality, and emergency care access; interventions which all have public health benefits.  For instance, it is possible that police may not be able to cover less dense and more remote locations of Balochistan as well as other areas, which may facilitate the reporting of major RTC with several fatalities only. The variations at district level in Punjab also pointed to similar influences as almost all of districts with relatively high RTF per crash were away from the resourceful provincial capital, Lahore. Available data, however, was limited in higher RTF per crash only to police resources and findings could also be inferred as either increasing severity of crashes by non-use of injury control measures e.g. seat-belts or saturation/unavailability of prehospital care, both factors being frequent problems in Pakistan. , The decrease in RTF per crash after 2004 in Punjab strongly favored prehospital care hypotheses as a government supported ambulance services had been operationalized in most of the centrally located districts in the province since 2004. , Clarifications are therefore needed about the factors associated with relatively higher RTF per crash at sub-national levels in Pakistan in order to appropriately prioritize public health interventions.
Relatedly, other RTF indicators increased in Balochistan as opposed to other provinces. Previous work based on state-wise data from India suggested that RTF rates had an inverse U relationship with gross national income (GNI).  In fact, the improvements in RTF indicators with increasing GNI over a certain limit indirectly reflect the development of safety promoting interventions such as effective police enforcement, newer vehicle fleet and high-quality road infrastructure.  Balochistan is the largest province of Pakistan by area with the lowest population density per square kilometer and is generally considered to be underdeveloped.  It is highly likely that increasing development indirectly contributed to higher RTF risks in this province, and therefore would require selective prevention efforts. ,
Overall, our findings advocate the need for improving surveillance of the RTC epidemic in Pakistan. Considering the available scenario, availability of district-wise RTC surveillance could help to better appreciate local RTC burdens at provincial as well as at district levels.  For instance, in 2006-2007, with investments of approximately 6 million Pakistani rupees (<100 000 US$), a RTC-specific surveillance system was able to collect data of over 35 thousand patients in Karachi city (Sindh).  The potential and feasibility of such systems were assessed in the district Rawalpindi (Punjab) as well.  Extending and financially supporting specific surveillance systems to other districts could permit to cost-effectively generate useful information for the policy makers at sub-national levels. 
Generalization of result: Important lessons for regional assessments
These results cannot be directly generalized to other regional countries like India and Bangladesh which have similar systems of police reporting.  Nonetheless, findings suggested that RTF trends from these regional LMICs should be interpreted considering sub-national or disaggregated data that might be useful in prioritizing interventions.
Our analyses have some limitations. Inclusion of explanatory variables such as GNI, hospital beds, registered vehicles and road kilometer per district, availability of police force, and ambulances available could provide a more comprehensive picture of sub-national road safety trends but were not available for analyses. Nonetheless, disparity in RTF data did indicate high risk areas for possible interventions at sub-national levels. ,
| Conclusions|| |
In view of the upcoming decade of action of road safety,  this exploratory study highlighted the limitations of actual RTF surveillance at sub-national levels in Pakistan, findings that might be of significance for other LMICs when analyzing their RTF data and prioritizing preventive interventions accordingly. This study showed that the RTF trends varied considerably at provincial and district levels in Pakistan during the last decade (2000-2009). It is likely that changing crash severity, healthcare access, or underreporting might be associated with these variations;  however, their relative contribution could not be determined without detailed investigations. Our findings suggested the need of strengthening police reporting or replacing the current reporting system by specific surveillance systems at all districts levels in order to improve RTC reporting in different Pakistani settings.
| Acknowledgements|| |
The authors are thankful to the three anonymous reviewers for their comments, which helped us to improve the interpretation and presentation of our findings. We are grateful to Samantha Wells, Douglas Hospital Research Centre for language editing and comments on presentation.
| References|| |
|1.||Peden M, Scurfiled R, Sleet D, Mohan D, Hyder A, Jarawan E. World report on road traffic injury prevention. Geneva: World Health Organization; 2004. |
|2.||Jacobs G, Aeron-Thomas A, Astrop A. Estimating global road fatalities. Crowthorne (UK): Transport Research Laboratory; 2000. |
|3.||Kopits E, Cropper M. Traffic fatalities and economic growth. Accid Anal Prev 2005;37:169-78. |
|4.||World Health Organization. Global Plan for the Decade of Action; for Road Safety 2011-2020. Geneva: World Health Organization, 2011:7-21. [Available from: http://www.who.int/roadsafety/decade_of_action/plan/english.pdf] [Lat acessed on 2011]. |
|5.||Bhatti JA, Salmi LR, Lagarde E, Razzak JA. Concordance between road mortality indicators in high-income and low- and middle-income countries. Traffic Inj Prev 2010;11:173-7. |
|6.||Evans L. Traffic Safety. Bloomfield Hills, MI: Science Serving Society; 2004. |
|7.||Sauerzapf V, Jones AP, Haynes R. The problems in determining international road mortality. Accid Anal Prev 2010;42:492-9. |
|8.||Garg N, Hyder AA. Exploring the relationship between development and road traffic injuries: A case study from India. Eur J Public Health 2006;16:487-91. |
|9.||La Torre G, Van Beeck E, Quaranta G, Mannocci A, Ricciardi W. Determinants of within-country variation in traffic accident mortality in Italy: A geographical analysis. Int J Health Geogr 2007;6:49. |
|10.||International Road Federation (IRF). World Road Statistics 2008 (Data 2001 to 2006). Geneva: International Road Federation; 2008. |
|11.||World Health Organization. Global status report on road safety. Geneva: World Health Organization; 2009. |
|12.||Bhatti JA, Salmi LR. Challenges in Evaluating the Decade of Action for Road safety in Developing Countries: A Survey of Traffic Fatality Reporting Capacity in the Eastern Mediterranean Region. Traffic Inj Prev 2012;13:422-6. |
|13.||Government of Pakistan. Pakistan Economic Survey 2010-11. Islamabad: Economic Advisor′s Wing, Finance Division, Government of Pakistan; 2011. |
|14.||Nishtar S, Mohamud KB, Razzak J, Ghaffar A, Ahmed A, Khan SA, et al. Injury prevention and control: National Action Plan for NCD Prevention, Control and Health Promotion in Pakistan. J Pak Med Assoc 2004;54(12 Suppl 3):S57-68. |
|15.||Hyder AA, Ghaffar AA, Sugerman DE, Masood TI, Ali L. Health and road transport in Pakistan. Public Health 2006;120:132-41. |
|16.||Razzak JA, Shamim MS, Mehmood A, Hussain SA, Ali MS, Jooma R. A successful model of Road traffic injury surveillance in a developing country: Process and Lessons Learnt. BMC Public Health 2012;12:357. |
|17.||Government of Balochistan. Development Statistics of Balochistan 2007. Quetta: Bureau of Statistics, Planning and Development Departments, Government of Balichistan, 2007. |
|18.||Government of NWFP. NWFP Development Statistics 2008. Peshawar: Bureau of Statistics, Planning and Development Department, Government of NWFP, 2008. |
|19.||Government of NWFP and FATA. FATA development Statistics 2008. Peshawar: Bureau of Statistics, Planning and Development Department, Government of NWFP and FATA, 2008. |
|20.||Government of Pakistan. Census report of Pakistan 1998. Islamabad: Population Census Organization, Statistics Division, Government of Pakistan, 2001. |
|21.||Government of Punjab. Punjab Development Statistics 2005. Lahore: Bureau of Statistics, Government of Punjab; 2005. |
|22.||Government of Sindh. Development Statistics 2006. Karachi: Bureau of Statistics, Government of Sindh; 2006. |
|23.||Fassihuddin. Police and policing in Pakistan. Pak J Criminol 2010;2:131-45. |
|24.||Bhatti JA, Razzak JA, Lagarde E, Salmi LR. Differences in police, ambulance, and emergency department reporting of traffic injuries on Karachi-Hala road, Pakistan. BMC Res Notes 2011;4:75. |
|25.||Khoso AK, Ekman DS, Bhatti JA. Comparison of Highway Crash Reporting in Pakistan with the World Health Organization Injury Surveillance Guidelines. Traffic Inj Prev 2011;12:279-82. |
|26.||Government of Pakistan. Pakistan Penal Code. Islamabad: Government of Pakistan; 2011. |
|27.||Ghaffar A, Hyder AA, Masud TI. The burden of road traffic injuries in developing countries: The 1 st national injury survey of Pakistan. Public Health 2004;118:211-7. |
|28.||Federal Bureau of Statistics. Pakistan Statistical Year Book 2010. Islamabad: Federal Bureau of Statistics, Statistics Division, Government of Pakistan; 2010. |
|29.||Waseem H, Naseer R, Razzak JA. Establishing a successful pre-hospital emergency service in a developing country: Experience from Rescue 1122 service in Pakistan. Emerg Med J 2011;28:513-5. |
|30.||Peden M. UN General Assembly calls for decade of action for road safety. Inj Prev 2010;16:213. |
|31.||Farooq U, Bhatti JA, Siddiq M, Majeed M, Malik N, Razzak JA, et al. Road traffic injuries in Rawalpindi city, Pakistan. East Mediterr Health J 2011;17:647-53. |
[Table 1], [Table 2], [Table 3]
|This article has been cited by|
||Possible doubling of road traffic injury burden in Pakistan: findings from a population-based survey in 2013–2014
| ||J.A. Bhatti,I. Rafique,A. Azad,M.A. Hamid,M.A.N. Saqib,H. Qureshi |
| ||Public Health. 2017; 150: 166 |
|[Pubmed] | [DOI]|
||Expected years of life lost through road traffic injuries in Mexico
| ||Efrén Murillo-Zamora,Oliver Mendoza-Cano,Benjamín Trujillo-Hernández,José Guzmán-Esquivel,Alfredo Medina-González,Miguel Huerta,Ramón Alberto Sánchez-Piña,Agustin Lugo-Radillo |
| ||Global Health Action. 2017; 10(1): 1360629 |
|[Pubmed] | [DOI]|