[Skip to Navigation]
Comment & Response
November 2018

Re-examining the Association Between “4/20” and Fatal Crashes—Doobie-ous Data?

Author Affiliations
  • 1Department of Surgery and Perioperative Care, Dell Medical School at the University of Texas, Austin
  • 2David Geffen School of Medicine at UCLA, Los Angeles, California
JAMA Intern Med. 2018;178(11):1565. doi:10.1001/jamainternmed.2018.5856

To the Editor In an analysis published in a recent issue of JAMA Internal Medicine, Staples and Redelmeier1 reported higher numbers of drivers involved in fatal motor vehicle crashes on April 20, the “counterculture holiday known as ‘4/20,’”1(p569) when compared with the corresponding day of the previous (April 13) and following (April 27) weeks. However, analyzing the number of drivers involved in crashes—rather than the number of crashes—inflates crash numbers (and absolute differences in comparative crash numbers) because the majority of crashes included in the Fatality Analysis Reporting System database2 involved more than 1 vehicle. We were also concerned that the authors appeared to have aggregated data for the 2 control dates, which can create the impression of an effect when one does not exist. For example, the number of crashes could be exactly the same on April 20 and on April 13, but a 20% lower crash count on April 27 would give the impression of a 10% excess in crashes attributable to 4/20. Finally, the analysis as presented did not fully convey the longitudinal nature of the data or more recent evolutions in any observed effect.

Add or change institution