Background Road traffic injuries are a major cause of death and disability in Africa. Effective action to prevent these deaths and injuries requires high-quality data on the total burden and characteristics of those affected. In Ghana, police crash records are the primary source of surveillance systems for road traffic deaths and injuries, however, underreporting is common, especially among vulnerable road users. Generating reliable estimates of deaths require an additional data source to be used with police crash records. In the absence of a robust national vital registration system in Ghana, hospital records are the only reliable complementary source.

Objective Collect road traffic fatality data from major referral hospitals to provide a basis for re-estimating road traffic mortality and describe the epidemiologic profile of casualties.

Methods A descriptive cross-sectional design approach was used to collect retrospective data in a major referral hospital in Kumasi, Ghana. Data abstraction was completed for cases of road injury which required hospital admission or resulted in death before admission for 2020 and 2021. Data on personal identifying variables and crash location were collected to facilitate linkage to police fatal crash records. Deaths from crashes outside the city’s boundary or in unknown locations were excluded from the linkage. A standard capture-recapture calculation using the Chapman estimator was applied to estimate road traffic fatalities in Kumasi. Capture-recapture is an approach for providing estimates of an event based on cases that are captured by more than one data source.

Results 132 deaths occurred from crashes in the city for the two-year period in hospital records. 59 of these deaths were linked to police records. The capture-recapture analysis estimated 663 road traffic deaths, contrasting with 294 police-reported deaths for the same period. This suggests that the re-estimated count is 2.3 times higher than police-reported deaths. The re-estimated counts were higher among pedestrians and motorcyclists compared to vehicle occupants.

Conclusions The findings reinforce the importance of linking multiple data sources to generate a more accurate estimate of road traffic deaths to inform planning and implementation of interventions in healthcare delivery, transport sector, and police enforcement to improve outcomes.