Aim To assess the relationship between risk perception and reported traffic collisions (TC) and traffic injuries (TI) the year before.

Methods Design: Cohorts.

Setting Primary Care (PC). Urban area. Barcelona (Spain).

Population 1938 subjects possessing driving license, who attended a participating practice from March to November/2009.

Measurements Outcome: TC, TI the year before.

Main exposures Perception of risk for TCs (self-perception (SPR) and as assessed by a health professional (HPPR)), scored 0–10.

Covariates Age, gender, seatbelt use, adherence to speed limits, long-term conditions (LTC) and drugs which increase risk for TC (LTD), psychoactive substance (PS).

Data collection Interview with a health professional during the recruitment visit. Checked with medical records.

Statistics Wilcoxon-test to assess differences in risk perception between categories. Logistic regression to assess relationship between risk perception scores, covariates and TC, TI.

Results Patients that take LTD scored SPR lower (mean difference (MD) -0.561(−0.938 to −0.183);p=0.019); those with LTC, and PS users dont score higher (p=0.09, p=0.35 respectively). Health professionals score higher those with LTC (MD 0.879(0.634 to 1.123); p<0.001), LTD (MD 0.967 (0.722 to 1.212); p<0.001) and PS users (MD 0.857 (0.497 to 1.216); p<0.001). HPPR is independently associated with TCs (p=0.026), and TIs (p=0.025). SPR is not related with TCs, nor it is with TIs (both p>0.5).

Discussion PC patients are not aware of the increased risk associated with their LTD, nor with the LTC they are prescribed; PS users arent either. By contrast, HPPRs know about the excess of risk associated with these. Their advice could be an effective intervention.