Background Recently in Great Britain, there has been a general decrease in the number of incidents resulting in injuries to bus passengers. However, in London the numbers of bus passenger casualties have increased since 2012, rising from a 31% share of the GB total in 2012 to 44% in 2016.

Methods Two datasets were analysed to explore the problem of bus passenger injury on London buses over a five-year period (2012–2016 inclusive); the STATS19 police collision data for Greater London and the Transport for London (TfL), Incident Reporting Information System (IRIS) containing bus operator reported passenger incident data.

Results In total 5,699 incidents were reported in STATS19 where at least one bus or coach passenger was injured and 7,043 reported casualties. In contrast the IRIS data reported 20,490 incidents and 21,998 injured passengers. In STATS19 non-collision incidents predominated (81%), and ‘slip, trip, fall’ was the main injury event (58%) in the IRIS data. STATS19 data classified injury severity as ‘slight’ (93%), ‘serious’ (7%) or ‘fatal’ (<0.1%). of the IRIS data only 6967 (32%) records contained useful injury description data, described as ‘minor’ or cuts, abrasions or bruises. However, 35% of the useful records reported injury mechanisms, e.g. ‘bump’ or ‘crush’ and not an injury descriptor.

Conclusion The lack of detailed injury information is a problem to understanding the severity of bus passenger injuries. High level severity information is collected in STATS19, but the IRIS dataset would benefit from better use of the ‘language’ of injury to differentiate description from mechanism.