- Email: [email protected]
Alcohol-related road traffic injury and Global Burden of Disease 2010
Correspondence
4
Kolias AG, Jones TL, Cowie CJ, et al. A report from the inaugural meeting of the British Neurosurgical Trainee Research Collaborative held in the Royal College of Surgeons of England, 19 October 2012. Br J Neurosurg 2013; 27: 307–10. Morton D. RCS to Expand Surgical Research Infrastructure. Ann R Coll Surg Engl (Suppl) 2012; 94: 131–32.
5
Alcohol-related road traffic injury and Global Burden of Disease 2010 Quantification of the disease burden of potentially modifiable risk factors is crucial for health decision-makers. The 2010 Global Burden of Disease (GBD) study provides an update of comparative risk assessment of the burden of disease and injury A
B
USA
25 000
GBD 2010 estimates Reported data
20 000 Number of deaths
attributable to 67 risk factors and cluster risk factors.1 However, the validity of GBD 2010 estimates of burden attributable to risk factors remains unreported because of the lack of country-level directly reported data. We analysed data for road injuries from the USA, Canada, UK, and New Zealand, and compared them with alcohol-related deaths from transport injury from GBD 2010. The reported alcohol-related deaths of these four countries were limited to road traffic injury and therefore should be less important than the GBD 2010 estimates for all alcoholrelated transport deaths. Yet, all four countries reported numbers of deaths that were notably larger than the GBD 2010 estimates except for 2010 in the UK (figure). This unexpected difference Canada
900 800 700 600
15 000
500 400
10 000
300 200
5 000
100 0
C
0
D
UK
800
We declare that we have no conflicts of interest. We acknowledge support from the 2009 New Century Scholar Support Program of the Ministry of Education of China (NCET-10-0782).
180
700 Number of deaths
New Zealand
200
160
600
140 500
120
*Guoqing Hu, Keita Mamady
400
100
[email protected]
300
80
Department of Epidemiology and Health Statistics, School of Public Health, Central South University, Changsha, 410078 Hunan, China
60
200
40 100
1
20
0
0 1990
1995
2000
2005
2010
1990
1995
2000
2005
2010
Figure: Transport-related deaths attributed to alcohol use in the USA, Canada, UK, and New Zealand, 1990–2010 (A) USA: based on the number of people killed by drivers with a blood alcohol concentration (BAC) ≥0·1 g/L.2 (B) Canada: based on the number of fatally injured drivers with BAC ≥1 mg%.3 (C) UK: based on the estimated number of persons killed by drink drive accidents BAC ≥80 mg%.4,5 (D) New Zealand: based on the estimated number of killed drivers with BAC over the legal limit (≥30 mg% from 1993).6
1092
could be primarily due to differences in estimation methods. Compared with the methods based on direct observation that the four countries used,2–6 the GBD 2010 used a five-step approach1 to estimate alcohol-related transport injury that was based on systematical review and synthesis of published and unpublished data because most countries lack highquality data such as the data for these four countries.The GBD method is more likely to be affected by some potential confounding factors than the direct method because few countries have data that can satisfy the requirement of the GBD five-step method. For example, the aetiological effect size of alcohol use on transport injury may change over time and across countries, as other factors change. Country-specific and yearspecific effect sizes of alcohol use are unavailable. The GBD group has to use a systematic review to estimate the effect size and apply it to all countries and all years. Because of the large difference between GBD data and reported data from four countries with high-quality data, the GBD 2010 data should be used cautiously to support policy related to alcohol use. When possible, high-quality directly observed data should be given priority, and further studies are needed to explain the large differences observed.
2
Lim SS, Vos T, Flaxman AD, et al. A comparative risk assessment of burden of disease and injury attributable to 67 risk factors and risk factor clusters in 21 regions, 1990–2010: a systematic analysis for the Global Burden of Disease Study 2010. Lancet 2012; 380: 2224–60. National Highway Traffic Safety Administration. Fatality analysis reporting system (FARS) encyclopedia. http://www-fars. nhtsa.dot.gov/Main/index.aspx (accessed July 15, 2013).
www.thelancet.com Vol 382 September 28, 2013
Correspondence
4
5
6
Traffic Injury Research Foundation of Canada. CCMTA road safety report series: alcohol-crash problem in Canada: 2010. http://tirf.ca/ publications/PDF_publications/2010_Alcohol_ Crash_Problem_Report_4_FINAL.pdf (accessed July 15, 2013). The Institute of Alcohol Studies. Drinking & driving: IAS factsheet, 2013. http://www.ias. org.uk/uploads/pdf/Factsheets/Drink%20 driving%20FS%20HM%20May%202013.pdf (accessed July 15, 2013). Department for Transport. Reported road casualties in Great Britain: 2010 annual report. http://assets.dft.gov.uk/statistics/releases/roadaccidents-and-safety-annual-report-2010/ rrcgb2010-00.pdf (accessed July 15, 2013). Ministry of Transport. Motor vehicle crashes in New Zealand 2011. http://www.transport.govt. nz/research/Pages/MotorVehicleCrashes inNewZealand2011.aspx (accessed July 15, 2013).
Authors’ reply We thank Guoqing Hu and Keita Mamady for their comments in which they present evidence from four countries on the number of people fatally injured by drivers with blood alcohol concentrations (BACs) above various thresholds, and compare these estimates to estimated alcoholattributable fatalities from the 2010 Global Burden of Disease (GBD) study.1 The BAC derived data indicate a higher attribution of traffic fatalities to alcohol use than was estimated in the GBD 2010. However, for traffic fatalities, a person having a BAC above a certain threshold does not indicate causality, nor does a person having a BAC below a certain threshold indicate that alcohol use did not have a causal role in the accident. For example, while alcohol use might affect reaction time and other cognitive processes for people with a BAC ≥0·03 g/L,2 this does not mean that all injury fatalities involving such BACs are caused by alcohol use. The same is true for higher levels of BAC, with a BAC ≥0·1 g/L not implying causality, even though this threshold has been used as an indicator of injuries caused by alcohol use. An alternative explanation for at least part of the association between alcohol use and injuries is that personality factors, such as sensation seeking, can affect both risky drinking and risky driving behaviour.3 When measuring the effect of alcohol on the risk of injury, www.thelancet.com Vol 382 September 28, 2013
such alternative explanations can only be excluded if personality factors are controlled for, for instance by use of risk relations from case-crossover studies when calculating alcohol-attributable fractions (see Goodarz Danaei and colleagues’ report4 for example). While not perfect, the approach used to calculate alcohol-attributable traffic injuries in the GBD 2010 is based on relative risks from more controlled studies.5 Thus, the current GBD methods are preferable when compared with an approach that bases causality solely on a BAC threshold; however, these methods lead to more conservative estimates. This is partly due to GBD methods relying on population survey data which underestimate the frequency of risky drinking occasions.5 Thus, for the next iteration of the GBD study, researchers should strive to further refine the methods used to estimate the number of alcohol-attributable road traffic injuries by correcting for the underestimation of survey-based exposure data as well as by including factors which may mediate the alcohol-injury relationship (such as road conditions). We declare that we have no conflicts of interest.
*Jürgen Rehm, Kevin D Shield, Theo Vos, Christopher J L Murray, Stephen Lim Centre for Addiction and Mental Health, Toronto, ON M5S 2S1, Canada (JR, KDS); Addiction Policy, Dalla Lana School of Public Health, University of Toronto, Toronto, ON, Canada (JR); Institute of Medical Science, University of Toronto, Toronto, ON, Canada (JR, KDS); and Institute for Health Metrics and Evaluation, University of Washington, Seattle, WA, USA (TV, CJLM, SL) 1
2
3
4
Lim SS, Vos T, Flaxman AD, et al. A comparative risk assessment of burden of disease and injury attributable to 67 risk factors and risk factor clusters in 21 regions, 1990-2010: a systematic analysis for the Global Burden of Disease Study 2010. Lancet 2012; 380: 2224–60. Eckardt M, File S, Gessa G, et al. Effects of moderate alcohol consumption on the central nervous system. Alcohol Clin Exp Res 1998; 22: 998–1040. Jonah BA. Sensation seeking and risky driving: a review and synthesis of the literature. Accid Anal Prev 1997; 29: 651–65. Danaei G, Ding E, Mozaffarian D, et al. The preventable causes of death in the United States: comparative risk assessment of lifestyle, dietary, and metabolic risk factors. PLoS Med 2009; 6: e1000058.
5
Shield KD, Gmel G Jr, Patra J, Rehm J. Global burden of injuries attributable to alcohol consumption in 2004: a novel way of calculating the burden of injuries attributable to alcohol consumption. Pop Health Metrics 2012; 10: 9.
Treatment for patients with indolent and mantle cell lymphoma Mathias Rummel and colleagues 1 (April 6, p 1203) report that median progression-free survival for patients with indolent lymphoma and mantle cell lymphoma was significantly improved with bendamustine plus rituximab compared with CHOP (cyclophosphamide, doxorubicin, vincristine, and prednisone) plus rituximab (R-CHOP). Furthermore, bendamustine plus rituximab was associated with fewer serious adverse events than was R-CHOP. However, several issues warrant further consideration. First, three cases with pulmonary embolism were documented in the bendamustine plus rituximab group, whereas no cases were reported in the R-CHOP group. Venous thromboembolism is the second leading cause of death in patients with cancer. Because patients with lymphoma receiving chemotherapy are at high risk of venous thromboembolism, we have concerns about whether there was an unbalanced distribution of risk factors related to venous thromboembolism between the two groups or if there is the possibility of increased risk of thromboembolic events attributable to the bendamustine plus rituximab combination therapy. Second, in current clinical practices, novel prognostic models are increasingly used, such as the mantle cell lymphoma international prognostic index and the follicular lymphoma international prognostic index 2 in the era of immunochemotherapy.2,3 We would be grateful
Simon Fraser/Science Photo Library
3
1093
4
Kolias AG, Jones TL, Cowie CJ, et al. A report from the inaugural meeting of the British Neurosurgical Trainee Research Collaborative held in the Royal College of Surgeons of England, 19 October 2012. Br J Neurosurg 2013; 27: 307–10. Morton D. RCS to Expand Surgical Research Infrastructure. Ann R Coll Surg Engl (Suppl) 2012; 94: 131–32.
5
Alcohol-related road traffic injury and Global Burden of Disease 2010 Quantification of the disease burden of potentially modifiable risk factors is crucial for health decision-makers. The 2010 Global Burden of Disease (GBD) study provides an update of comparative risk assessment of the burden of disease and injury A
B
USA
25 000
GBD 2010 estimates Reported data
20 000 Number of deaths
attributable to 67 risk factors and cluster risk factors.1 However, the validity of GBD 2010 estimates of burden attributable to risk factors remains unreported because of the lack of country-level directly reported data. We analysed data for road injuries from the USA, Canada, UK, and New Zealand, and compared them with alcohol-related deaths from transport injury from GBD 2010. The reported alcohol-related deaths of these four countries were limited to road traffic injury and therefore should be less important than the GBD 2010 estimates for all alcoholrelated transport deaths. Yet, all four countries reported numbers of deaths that were notably larger than the GBD 2010 estimates except for 2010 in the UK (figure). This unexpected difference Canada
900 800 700 600
15 000
500 400
10 000
300 200
5 000
100 0
C
0
D
UK
800
We declare that we have no conflicts of interest. We acknowledge support from the 2009 New Century Scholar Support Program of the Ministry of Education of China (NCET-10-0782).
180
700 Number of deaths
New Zealand
200
160
600
140 500
120
*Guoqing Hu, Keita Mamady
400
100
[email protected]
300
80
Department of Epidemiology and Health Statistics, School of Public Health, Central South University, Changsha, 410078 Hunan, China
60
200
40 100
1
20
0
0 1990
1995
2000
2005
2010
1990
1995
2000
2005
2010
Figure: Transport-related deaths attributed to alcohol use in the USA, Canada, UK, and New Zealand, 1990–2010 (A) USA: based on the number of people killed by drivers with a blood alcohol concentration (BAC) ≥0·1 g/L.2 (B) Canada: based on the number of fatally injured drivers with BAC ≥1 mg%.3 (C) UK: based on the estimated number of persons killed by drink drive accidents BAC ≥80 mg%.4,5 (D) New Zealand: based on the estimated number of killed drivers with BAC over the legal limit (≥30 mg% from 1993).6
1092
could be primarily due to differences in estimation methods. Compared with the methods based on direct observation that the four countries used,2–6 the GBD 2010 used a five-step approach1 to estimate alcohol-related transport injury that was based on systematical review and synthesis of published and unpublished data because most countries lack highquality data such as the data for these four countries.The GBD method is more likely to be affected by some potential confounding factors than the direct method because few countries have data that can satisfy the requirement of the GBD five-step method. For example, the aetiological effect size of alcohol use on transport injury may change over time and across countries, as other factors change. Country-specific and yearspecific effect sizes of alcohol use are unavailable. The GBD group has to use a systematic review to estimate the effect size and apply it to all countries and all years. Because of the large difference between GBD data and reported data from four countries with high-quality data, the GBD 2010 data should be used cautiously to support policy related to alcohol use. When possible, high-quality directly observed data should be given priority, and further studies are needed to explain the large differences observed.
2
Lim SS, Vos T, Flaxman AD, et al. A comparative risk assessment of burden of disease and injury attributable to 67 risk factors and risk factor clusters in 21 regions, 1990–2010: a systematic analysis for the Global Burden of Disease Study 2010. Lancet 2012; 380: 2224–60. National Highway Traffic Safety Administration. Fatality analysis reporting system (FARS) encyclopedia. http://www-fars. nhtsa.dot.gov/Main/index.aspx (accessed July 15, 2013).
www.thelancet.com Vol 382 September 28, 2013
Correspondence
4
5
6
Traffic Injury Research Foundation of Canada. CCMTA road safety report series: alcohol-crash problem in Canada: 2010. http://tirf.ca/ publications/PDF_publications/2010_Alcohol_ Crash_Problem_Report_4_FINAL.pdf (accessed July 15, 2013). The Institute of Alcohol Studies. Drinking & driving: IAS factsheet, 2013. http://www.ias. org.uk/uploads/pdf/Factsheets/Drink%20 driving%20FS%20HM%20May%202013.pdf (accessed July 15, 2013). Department for Transport. Reported road casualties in Great Britain: 2010 annual report. http://assets.dft.gov.uk/statistics/releases/roadaccidents-and-safety-annual-report-2010/ rrcgb2010-00.pdf (accessed July 15, 2013). Ministry of Transport. Motor vehicle crashes in New Zealand 2011. http://www.transport.govt. nz/research/Pages/MotorVehicleCrashes inNewZealand2011.aspx (accessed July 15, 2013).
Authors’ reply We thank Guoqing Hu and Keita Mamady for their comments in which they present evidence from four countries on the number of people fatally injured by drivers with blood alcohol concentrations (BACs) above various thresholds, and compare these estimates to estimated alcoholattributable fatalities from the 2010 Global Burden of Disease (GBD) study.1 The BAC derived data indicate a higher attribution of traffic fatalities to alcohol use than was estimated in the GBD 2010. However, for traffic fatalities, a person having a BAC above a certain threshold does not indicate causality, nor does a person having a BAC below a certain threshold indicate that alcohol use did not have a causal role in the accident. For example, while alcohol use might affect reaction time and other cognitive processes for people with a BAC ≥0·03 g/L,2 this does not mean that all injury fatalities involving such BACs are caused by alcohol use. The same is true for higher levels of BAC, with a BAC ≥0·1 g/L not implying causality, even though this threshold has been used as an indicator of injuries caused by alcohol use. An alternative explanation for at least part of the association between alcohol use and injuries is that personality factors, such as sensation seeking, can affect both risky drinking and risky driving behaviour.3 When measuring the effect of alcohol on the risk of injury, www.thelancet.com Vol 382 September 28, 2013
such alternative explanations can only be excluded if personality factors are controlled for, for instance by use of risk relations from case-crossover studies when calculating alcohol-attributable fractions (see Goodarz Danaei and colleagues’ report4 for example). While not perfect, the approach used to calculate alcohol-attributable traffic injuries in the GBD 2010 is based on relative risks from more controlled studies.5 Thus, the current GBD methods are preferable when compared with an approach that bases causality solely on a BAC threshold; however, these methods lead to more conservative estimates. This is partly due to GBD methods relying on population survey data which underestimate the frequency of risky drinking occasions.5 Thus, for the next iteration of the GBD study, researchers should strive to further refine the methods used to estimate the number of alcohol-attributable road traffic injuries by correcting for the underestimation of survey-based exposure data as well as by including factors which may mediate the alcohol-injury relationship (such as road conditions). We declare that we have no conflicts of interest.
*Jürgen Rehm, Kevin D Shield, Theo Vos, Christopher J L Murray, Stephen Lim Centre for Addiction and Mental Health, Toronto, ON M5S 2S1, Canada (JR, KDS); Addiction Policy, Dalla Lana School of Public Health, University of Toronto, Toronto, ON, Canada (JR); Institute of Medical Science, University of Toronto, Toronto, ON, Canada (JR, KDS); and Institute for Health Metrics and Evaluation, University of Washington, Seattle, WA, USA (TV, CJLM, SL) 1
2
3
4
Lim SS, Vos T, Flaxman AD, et al. A comparative risk assessment of burden of disease and injury attributable to 67 risk factors and risk factor clusters in 21 regions, 1990-2010: a systematic analysis for the Global Burden of Disease Study 2010. Lancet 2012; 380: 2224–60. Eckardt M, File S, Gessa G, et al. Effects of moderate alcohol consumption on the central nervous system. Alcohol Clin Exp Res 1998; 22: 998–1040. Jonah BA. Sensation seeking and risky driving: a review and synthesis of the literature. Accid Anal Prev 1997; 29: 651–65. Danaei G, Ding E, Mozaffarian D, et al. The preventable causes of death in the United States: comparative risk assessment of lifestyle, dietary, and metabolic risk factors. PLoS Med 2009; 6: e1000058.
5
Shield KD, Gmel G Jr, Patra J, Rehm J. Global burden of injuries attributable to alcohol consumption in 2004: a novel way of calculating the burden of injuries attributable to alcohol consumption. Pop Health Metrics 2012; 10: 9.
Treatment for patients with indolent and mantle cell lymphoma Mathias Rummel and colleagues 1 (April 6, p 1203) report that median progression-free survival for patients with indolent lymphoma and mantle cell lymphoma was significantly improved with bendamustine plus rituximab compared with CHOP (cyclophosphamide, doxorubicin, vincristine, and prednisone) plus rituximab (R-CHOP). Furthermore, bendamustine plus rituximab was associated with fewer serious adverse events than was R-CHOP. However, several issues warrant further consideration. First, three cases with pulmonary embolism were documented in the bendamustine plus rituximab group, whereas no cases were reported in the R-CHOP group. Venous thromboembolism is the second leading cause of death in patients with cancer. Because patients with lymphoma receiving chemotherapy are at high risk of venous thromboembolism, we have concerns about whether there was an unbalanced distribution of risk factors related to venous thromboembolism between the two groups or if there is the possibility of increased risk of thromboembolic events attributable to the bendamustine plus rituximab combination therapy. Second, in current clinical practices, novel prognostic models are increasingly used, such as the mantle cell lymphoma international prognostic index and the follicular lymphoma international prognostic index 2 in the era of immunochemotherapy.2,3 We would be grateful
Simon Fraser/Science Photo Library
3
1093