|
Affiliation
|
Department of Public Health Sciences, University of Chicago, Chicago, IL 60637, USA. Electronic address: kavibhalla@gmail.com. |
|
Abstract
|
We need reliable statistics of road traffic injuries for two reasons: first, when such statistics enable comparisons with other health conditions, they allow traffic injuries to be prioritised appropriately on the national policy agenda. Second, detailed and timely statistics provide evidence of the effectiveness of safety measures. The estimates from the Global Burden of Diseases, Risk Factors, and Injuries Study (GBD) have proven extremely valuable in establishing that traffic injuries are a pressing societal concern in low-income and middle-income countries; however, the large uncertainties in GBD estimates should not be overlooked.
In The Lancet Public Health, the India State-Level Disease Burden Initiative Road Injury Collaborators1 present estimates of road injuries in India from GBD 2017; the implications of the uncertainty in these estimates need to be understood. The key underlying source of mortality data for this study was the national Sample Registration System, which includes verbal autopsy data on cause of death in a state-level representative population.2 GBD 2017 used Sample Registration System data for 2004–13 to estimate 219 000 deaths due to road injuries in India in 2017.1 Although the estimate appears to have a high level of certainty (95% uncertainty interval [UI] 202 000–231 000), comparisons with other analyses that used the same primary data (Sample Registration System 2004–13) reveal that uncertainty is much higher. Menon and colleagues3 estimated 275 000 deaths due to road injuries (95% UI not reported) in 2017, 26% higher than that reported in GBD 2017.1 These varying estimates suggest that the uncertainty in GBD's estimate could be about three times that reported—ie, about 20% higher or lower at the national level, and much higher at the state level where verbal autopsy sample sizes are much smaller. Notably, the primary underlying data are the same in these analyses, and so much of the uncertainty could be due to variations in GBD modelling strategies that are unaccounted for and which are unclear to an external audience ...
Language: en |