Background Unlike the UK or New Zealand, there is no standard set of census variables in the USA for characterising socioeconomic (SES, socioeconomic status) inequalities in health outcomes, including injury. We systematically reviewed existing US studies to identify conceptual and methodological strengths and limitations of current approaches to determine those most suitable for research and surveillance.

Methods We searched seven electronic databases to identify census variables proposed in the peer-reviewed literature to monitor injury risk. Inclusion criteria were that numerator data were derived from hospital, trauma or vital statistics registries and that exposure variables included census SES constructs.

Results From 33 eligible studies, we identified 70 different census constructs for monitoring injury risk. Of these, fewer than half were replicated by other studies or against other causes, making the majority of studies non-comparable. When evaluated for a statistically significant relationship with a cause of injury, 74% of all constructs were predictive of injury risk when assessed in pairwise comparisons, whereas 98% of all constructs were significant when aggregated into composite indices. Fewer than 30% of studies selected SES constructs based on known associations with injury risk.

Conclusions There is heterogeneity in the conceptual and methodological approaches for using census data for monitoring injury risk as well as in the recommendations as to how these constructs can be used for injury prevention. We recommend four priority areas for research to facilitate a more unified approach towards use of the census for monitoring socioeconomic inequalities in injury risk.