San Francisco pedestrian injury surveillance: mapping, under-reporting, and injury severity in police and hospital records

Accid Anal Prev. 2005 Nov;37(6):1102-13. doi: 10.1016/j.aap.2005.06.010. Epub 2005 Aug 8.


Goals: Police reports of severely injured pedestrians help identify hazardous traffic areas in San Francisco, but they under-report non-fatal collisions. We set out to: identify injured pedestrians who were missing from police collision reports, see what biases exist in injury reporting and assess the utility of broad categories of police severe injury (including fatal) for mapping and analysis.

Methods: We linked data on injured pedestrians from police collision reports listed in the Statewide Integrated Traffic Reporting System (SWITRS, n = 1991) with records of pedestrians treated at San Francisco General Hospital (SFGH, n = 1323) for 2000 and 2001. Data were analyzed using bivariate statistics, logistic regression and mapping.

Results: : We found that police collision reports underestimated the number of injured pedestrians by 21% (531/2442). Pedestrians treated at SFGH who were African-American were less likely then whites (odds ratio = 0.55, p-value < or= 0.01), and females were more likely than males (odds ratio = 1.5, p-value < or = 0.01) to have a police collision report. Over 70% of pedestrians deemed by the police to have a severe injury received treatment at SFGH, regardless of the collision's distance from SFGH. The sensitivity of a police-designated severe injury (including fatal) was 69% and the specificity was 89% when compared with a known SFGH assessment. But, sensitivity declined when we included pedestrians without a SFGH record.

Conclusion: Though collision reports have demonstrated limitations, broad categories of police severity may be sensitive enough to map locations where numerous severe injuries occur, for timely countermeasure selection.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Accidents, Traffic / classification
  • Accidents, Traffic / statistics & numerical data*
  • Adolescent
  • Adult
  • Aged
  • Bias
  • Child
  • Child, Preschool
  • Ethnicity / statistics & numerical data
  • Female
  • Hospitals, Urban / statistics & numerical data*
  • Humans
  • Infant
  • Male
  • Middle Aged
  • Police
  • Population Surveillance / methods*
  • Records
  • Risk Assessment
  • Risk Factors
  • San Francisco / epidemiology
  • Walking / injuries*