Close the road-death data gap before Bangladesh can cut its 5-8K annual toll

Diagnosis

Road-traffic injury is a tier-1 structural hazard, and the country cannot yet say with confidence how large it is. The curated estimate is 5-8K deaths per year, but that range itself is the problem: BRTA administrative counts and media-based counts diverge, so the floor and ceiling differ by thousands of lives. A policy system that does not know whether it is losing five thousand or eight thousand people a year cannot target enforcement, cannot evaluate whether any intervention worked, and cannot defend a budget line. The divergence is not a rounding artifact, it is a sign that fatalities are captured through two unreconciled channels (police and transport-authority records on one side, press reporting on the other), with no shared case definition or linkage to hospital and death-registration data. Lead responsibility sits with the Directorate General of Health Services (DGHS), with the Department of Public Health Engineering as a supporting body. That placement matters: it frames road death as a public-health surveillance and prevention problem, not only a traffic-policing one, which is the correct frame for closing the count gap and acting on it.

Recommended actions

1. Reconcile the two counts into one official figure. Owner: DGHS. Mechanism: a standing road-traffic-injury surveillance unit that links hospital injury records, the death-registration system, and BRTA crash records under a single WHO-style case definition, publishing one reconciled annual fatality number with its uncertainty band. Signal that it is working: the gap between the administrative count and the media-based count narrows toward a single reported figure with a documented method.
2. Stand up sentinel hospital injury surveillance. Owner: DGHS. Mechanism: designate trauma-receiving hospitals as sentinel reporting sites that log every road-traffic injury admission and outcome on a common form, feeding the surveillance unit monthly. Signal: a rising share of the 5-8K estimated deaths is captured as individual, verifiable cases rather than inferred from press reports.
3. Convert the data into a ranked blackspot and risk map. Owner: DGHS, with the Department of Public Health Engineering. Mechanism: geolocate reconciled crash and injury records to identify the corridors and intersections carrying the heaviest toll, and hand that list to road-design and engineering owners. Signal: a published, periodically updated blackspot list that names specific locations.
4. Tie remedial engineering to the blackspots. Owner: Department of Public Health Engineering (supporting DGHS). Mechanism: a funded works programme that treats the top-ranked locations first (junction redesign, pedestrian crossings, speed-calming, lighting), sequenced by the ranked list rather than by ad hoc request. Signal: treated blackspots show fewer subsequent injuries in the sentinel data.
5. Publish a quarterly public dashboard. Owner: DGHS. Mechanism: a routine release of the reconciled count, trend, and blackspot status, so the figure stops being contested. Signal: media, BRTA, and the surveillance unit converge on the same number in public discussion.

Sequencing (first 12 months)

Start with reconciliation and sentinel surveillance (actions 1 and 2): nothing else can be targeted or evaluated until there is one trusted count. DGHS can launch these on existing hospital and registration infrastructure without new legislation. Once the first reconciled quarters exist, the blackspot map (action 3) becomes possible, which in turn unlocks the engineering programme (action 4) and the public dashboard (action 5). The first year buys the measurement foundation; the toll-reduction work is sequenced off it.

Risks and constraints

The binding constraint is institutional, not technical: a credible single count can expose that the true toll sits at the upper end of the 5-8K range, which is politically uncomfortable and creates pressure to keep the channels unreconciled. Cross-agency data sharing between DGHS, hospitals, registration, and BRTA is the usual failure point, and the supporting engineering body competes for capital budget against many other claims. Sustained funding for the surveillance unit, not a one-off survey, is the fiscal risk.

Bottom line

Bangladesh loses an estimated 5-8K people a year on its roads, but the divergence between BRTA and media counts means the country is acting without a reliable number. DGHS should first reconcile the count and stand up hospital surveillance, then use the resulting blackspot map to direct engineering fixes, because every later intervention depends on first knowing the true toll.