2023 Public Health at BMJ

Short report

Racial inequity in fatal US police shootings, 2015 – 2020 Elle Lett , 1,2,3 Emmanuella Ngozi Asabor , 4,5 Theodore Corbin , 6 Dowin Boatright 7

1 Biostatistics, Epidemiology, and Informatics, University of Pennsylvania, Philadelphia, Pennsylvania, USA 2 Leonard Davis Institute of Health Economics, University of Pennsylvania, Philadelphia, PA, USA 3 Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA, USA 4 Yale University School of Medicine, New Haven, Connecticut, USA 5 Department of Epidemiology of Microbial Diseases, Yale School of Public Health, New Haven, CT, USA 6 Department of Emergency Medicine, Drexel University College of Medicine, Philadelphia, Pennsylvania, USA 7 Department of Emergency Medicine, Yale University School of Medicine, New Haven, CT, USA Correspondence to Elle Lett, Department of Biostatistics, Epidemiology, and Informatics, 423 Guardian Drive, Philadelphia, PA 19104, USA; lanair.lett@ pennmedicine.upenn.edu Received 11 July 2020 Revised 28 September 2020 Accepted 29 September 2020

ABSTRACT Introduction Violent encounters with police represent a signi fi cant cause of morbidity and mortality in the USA, especially among Black, Indigenous, and People of Colour (BIPOC). This study characterises trends in fatal police shootings overall and by armed status and quanti fi es inequities in mortality burden and years of life lost (YLL) across racial/ethnic groups. Methods Longitudinal study of Washington Post data on fatal police shootings in the USA using generalised linear-mixed models to capture trends with time and relative rates. Results This study shows that the rate of fatal police shootings for Black, Indigenous, and People of Colour (BIPOC) is constant from 2015 to 2020. Further, BIPOC have signi fi cantly higher death rates compared with Whites in the overall victim pool (Native American RR=3.06, Black RR=2.62, Hispanic RR=1.29) and among unarmed victims (Black RR=3.18, Hispanic RR=1.45). Native American (RR=3.95), Black (overall RR=3.29, unarmed RR=3.49) and Hispanic (RR=1.55, unarmed RR=1.55), victims had similarly high rates of YLL relative to Whites. Conclusion Fatal police shootings are a public health emergency that contribute to poor health for BIPOC. Urgent attention from health professionals is needed to help drive policy efforts that reduce this unjust burden and move us towards achieving health equity in the US. INTRODUCTION Interactions with police are an important cause of morbidity/mortality, particularly for Black, Indigenous, and People of Colour (BIPOC) in the USA. Since the fatal police shooting of Michael Brown, an unarmed Black man in Ferguson, MO, police violence has received increased societal scru- tiny. Previous studies have demonstrated that BIPOC experience a disproportionate mortality burden due to police violence, both in terms of net fatalities 1 and years of life lost. 2 Nevertheless, little is known about how the rate of police killings of BIPOC has changed over time. To address this critical knowledge gap, we examined overall and racial/ethnic group trends in fatal police shootings from 2015 to 2020, with additional attention direc- ted at the rate at which unarmed individuals are killed by the police. We also quantify the dispropor- tionate burden of fatal police shootings on BIPOC death rates and years of life lost during the study period.

METHODS Data

This study uses publicly available data on fatal police shootings provided by the Washington Post . The repository aims to collect information on every person killed by on-duty police officers in the USA. 3 Their methodology includes monitoring local news reports, independent databases and addi- tional reporting handled by the Post. Their data include victim race, age and sex, as well as details about any item in the victim ’ s possession that was perceived as a weapon. Two independent coders categorised the weapons into the following groups: blunt, firearm, knife, sharp (non-knife), stun, other, and none (unarmed), and reviewed discrepancies to reach an agreement. Tools, toys and debris were categorised as unarmed because of the low likeli- hood of immediate lethal force. Examples of items possessed by victims that were not treated as weap- ons in our analysis were ‘ air conditioner ’ , ‘ chair ’ , ‘ wasp spray ’ , ‘ pen ’ , and ‘ shovel ’ . Trend analysis We calculate the death rate and years of life lost (YLL) for all fatal police shootings per quarter per million (pqpm) from 2015 to the first quarter of 2020, and for fatal police shootings with an unarmed victim per half-year per million (phpm), from 2015 to 2019, by race/ethnic group. YLL is a summary measure of premature deaths that gives greater weight to death at a younger age. Denominators for rates are racial/ethnic group US population size estimates from the American Community Survey (ACS) for 2015 – 2018 with lin- early extrapolated estimates for 2019 and 2020, for which ACS data are unavailable. 4 We compare longitudinal trends using negative binomial general- ised linear-mixed models (GLMMs) with fixed and random effects for both race and time. 5 Rate ratios (RR) and 95% CIs are provided for Asian, Black, Hispanic and Native American populations relative to the White population for all fatal police shoot- ings. CIs that do not span 1 indicate statistical sig- nificance at the type I error rate of α =0.05. Asian and Native Americans are excluded from the unarmed victim trend analysis because of low counts. We estimate YLL based on the life expec- tancy for US citizens in victim birth year as the difference between life expectancy and age at death, consistent with previous studies. 2 Data on US life expectancy by birth year were obtained from statista based on historical data from the UN Department of Economic and Social Affairs. 6

To cite: Lett E, Asabor EN, Corbin T, et al. J Epidemiol Community Health Epub ahead of print: [ please include Day Month Year]. doi:10.1136/ jech-2020-215097

Lett E, et al. J Epidemiol Community Health 2020; 0 :1 – 4. doi:10.1136/jech-2020-215097