BACKGROUND

The police force in the USA has a long history of using excessive force against racial minorities, particularly blacks.¹ The public’s response to this injustice has ranged from riots in Los Angeles after the acquittal of the police officers charged with killing Rodney King in 1991 to protests and the creation of the Black Lives Matter (BLM) after the acquittal of the Trayvon Martin’s killer in 2013. Recently, the death of George Floyd in May 2020 further transformed this movement to a global protest.² Despite this ongoing crisis, little is known about the public health impacts of law enforcement-related deaths of unarmed black Americans.

These deaths might lead to collective trauma in the black community. Collective trauma refers to a psychological reaction to a traumatic event that affects a community and ‘results from human behaviour that is politically motivated and has political consequences. Such an event injures in one sharp state, penetrating all psychological defensive barriers of participants and observers, allowing no space for denial mechanisms and thus leaving those affected with an acute sense of vulnerability and fragility’.³ Given the history of race-based traumatic events perpetuated by those in positions of power in the USA, people of colour might be affected when learning about law enforcement-related deaths of unarmed victims in their community.

Law enforcement-related deaths of unarmed black Americans reflect the larger system of structural racism, defined as the systems in which public policies, institutional practices, cultural representations and other norms work in various, often reinforcing ways to perpetuate racial inequities in the USA.^{4 5} Experiences of interpersonal racism increase the odds of delayed healthcare or unmet medical need,^{6 7} exacerbating health disparities if increased mistrust is coupled with underutilisation of emergency healthcare services. Aggressive policing practices and police maltreatment are also associated with adverse health outcomes among populations that are most affected.^8–10 Furthermore, an emerging body of research has found an association between law enforcement-related deaths and increased mental health symptoms and self-reported poor mental health among black Americans.^{11 12}

Aggressive policing of black Americans is symptomatic of injustice,¹³ and might lead to institutional mistrust. Institutional mistrust is the perception of the unfairness and ineffectiveness of establishment in charge or position.¹⁴ Institutional mistrust may result if communities believe public institutions have not fairly fulfilled their social responsibilities.¹⁵ For example, the widespread disclosure of the Tuskegee study led to increased medical mistrust, decreased healthcare utilisation and increased mortality for older black men.¹⁶ Other studies have shown that greater mistrust for healthcare providers among black Americans is associated with the underutilisation of healthcare services.^{17 18} Separately, another study found a decrease in crime-related 911 calls from predominantly black communities after a well-publicised case of police violence against an unarmed black man.¹⁹

Institutional betrayal of social trust could lead to cross-institutional avoidance. We hypothesise that law enforcement-related deaths of unarmed black Americans will lead to mistrust of other public institutions such as hospital emergency departments (ED). Specifically, we tested two hypotheses: (1) ED utilisation among black New York residents will decrease immediately after the public becomes aware of a law enforcement-related death of an unarmed black New Yorker; (2) this effect will be modified by the larger societal context (ie, implementation of policies and programmes promoting police accountability, different political climate). We further explored whether any changes in ED utilisation are specific to mental health-related or ambulatory care-sensitive conditions (ACS).

RESULTS

In 2013–2016, there were 14 law enforcement-related deaths in NYS (6 in NYC) that met our eligibility criteria (figure 1). Twelve were reported by the media within 1 day of death. The annual number of incidents ranged from two in 2013 to seven in 2015. On average, ED rates were consistently higher for black compared with white New Yorkers. In 2013–2014, there was a sharp discontinuity in the ED rates for blacks immediately after a media report of on law enforcement-related deaths and a smaller discontinuity noted in 2015–2016 (figure 2).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Timeline of media report dates for law enforcement-related deaths of unarmed black New Yorkers, 2013–2016.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

A vertical line at day 0 indicates the day the press story of law enforcement-related death of an unarmed black person in New York State. Dots are the daily race-specific ED rates per 100,000 persons. We used nonparametric locally weighted smoothing to fit the trend line in the ED rates separately before and after a law enforcement-related death stratified by race.

Table 1 compares the race-specific results stratified by time period. Table 1 also summarises the differences in the results when using the date of media reporting, date of death or randomly generated dates as the cut-off (n=730 for 2013–2014). Using the date of the first media report for the cut-off and a 2-week exposure window, we found a decrease of 7.7 visits per 100 000 black New Yorkers (5% less than the daily average ED rate of 145 visits per 100 000 among black New Yorkers) in 2013–2014. Results were similar when we varied the time window (eg, decrease of 7.9 visits per 100 000 among black New Yorkers using a 4-week window). We found a smaller, non-significant decrease using the date of death as the cut-off (eg, decrease of 5.4 visits per 100 000 black New Yorkers). During 2015–2016, overall ED rates did not significantly change after a law enforcement-related death of an unarmed black New Yorker. Among white New Yorkers, the ED rate did not change using any model specification. Sensitivity analyses using randomly chosen dates did not yield statistically significant results for any of the race stratified models.

In subanalyses, we found no statistically significant difference in ACS ED rate for blacks or whites in 2014–2015 or in 2015–2016. Model results for mental health-related ED rates also did not differ by race or model specifications (online appendix tables 2–3). The principal diagnoses for black and white New Yorkers were similar before and after a law enforcement-related death in 2015–2016 (results not shown).

DISCUSSION

Our study found a decrease in the number of ED visits by black New Yorkers following the media report of law enforcement-related deaths of unarmed black New Yorkers during 2013–2014. In contrast, we did not observe a similar decrease among black in 2015–2016, nor for white New Yorkers during either time period. We found smaller estimates when we used the death instead of the media report date. The difference between using the date of death compared with the date of the first media report suggests knowledge diffusion of the deaths leads to community-level effects. Our results examining ACS-specific ED rates and mental health-related ED rates were not statistically significant for any models.

One possible explanation for the difference ED rates for black New Yorkers between these two time periods is the increased emphasis on police accountability in policy making and public discourse in 2015. Although there was rising activism in 2013, the community mobilisation needed to address police reforms did not gain momentum until 2015.³⁰ By 2015, policies and programmes addressing police accountability were implemented in the federal, state and local levels. Nationally, the Death in Custody Reporting Act (HR 1447) which required states who receive certain federal funding to provide information about the death of any person who is detained, arrested, en route to incarceration, or incarcerated in state or local facilities or at a boot camp prison to the Attorney General. It also included the National Initiative for Building Community Trust and Justice, a programme designed to improve relationships between law enforcement and communities of colour. On the state level, police accountability initiatives implemented in 2015 include the New York State Executive Order 8.147 which appointed the state’s attorney general as special prosecutor in cases where law enforcement officers were involved in the deaths of civilians. Locally, the New York City Police Department implemented three programmes starting in 2015: a neighbourhood policing programme where neighbourhood coordination officers serve as a liaison between the police and the community to strengthen and better address community concerns,³⁰ a pilot programme for police officers to wear body cameras, and an annual 3-day training on the use of force.³¹ Initiatives, such as those described above, might improve community trust in police. According to national polls, Americans’ reported confidence in the police decreased from 57% in 2013, to 53% in 2014 and 52% in 2015 before rebounded to 56% in 2016 with whites consistently reporting more trust in the police than blacks.^{32 33} A recent qualitative study also found the public’s perception of how socially equitable their local police department was increased when police departments had community policing programmes that increase citizen involvement and required staff to undergo more ethics training.³⁴ Further research is needed to understand whether effective police accountability policies and programmes may lead communities to greater trust in social institutions even in the event of community tragedies such as law enforcement-related deaths.

Other possible explanations for the difference in results for black ED rates between these two time periods is habituation as police killings of unarmed black individuals became more frequent in 2015–2016. Previous studies have shown that African American youth exposed to high levels of chronic community violence may become desensitised to violence.^{35 36} The lack of statistically significant changes in ED visits might also partially be a methodological artefact. The shortened time span between deaths in 2015 might preclude our ability to detect any sharp, immediate effect.

Our subanalyses examining mental health and ACS-related ED rates suggest the decrease in the overall rates was not driven by decrease in these specific ED rates. While a previous study found worst self-reported mental health among black Americans after law enforcement-related deaths,¹² we did not note any changes in mental health-related ED rates. This is potentially problematic if there is an increased need for mental healthcare among black Americans after law enforcement-related deaths but there is no change in mental health utilisation. The lack of statistical significance in our ACS-specific ED rates and mental health-related ED rates for any models may be because of lack of statistical power. We also note the change from ICD-9 to ICD-10 coding may have especially affected the results for the ACS-specific conditions because there may be additional ICD-10 codes for these conditions that were not captured in our analysis. More studies examining specific conditions may be helpful.

Further research is also needed to better understand possible longer-term community-level impact of law enforcement-related deaths. While our analysis estimated immediate changes in ED rates, we did not observe any long-term decreases in ED rates in exploratory trend analyses (ie, descriptive statistics and trend lines showed no long-term decrease in ED rates for black New Yorkers).

There are several limitations to our study. The MPV database might not have included all law enforcement-related deaths. The impact of additional, unreported events is unknown, and we did not measure the impact of events occurring outside of NYS. However, these findings suggest that the presence of other events would likely have biased these results towards the null. Although we hypothesised an immediate effect in ED-seeking behaviour following such deaths, there might be a lagged effect not captured in our analysis. Our model includes an indicator variable for whether the first media report of a law enforcement-related death occurred in the previous week. Because we used administrative claims data, there might be misclassification of mental health principal diagnosis and race. However, such misclassification should be random and unrelated to timing of law enforcement-related deaths incidents. There are variations in how race is ascertained and categorised in each hospital facility. However, a recent analysis found a relatively high sensitivity and positive predictive value for SPARCS compared with NYC birth certificates for non-Latino white and black race categories.³⁷ SPARCS transitioned from the use of ICD-9 to ICD-10 codes during September–October 2015; this is unlikely to have biased our results because none of the exposure periods for the individual incidents overlapped the transition period. Finally, our analysis has limited generalisability because it focused only on limited number of clearly documented incidents in NYS and does not extend to other healthcare settings.

Law enforcement-related deaths of unarmed black Americans reflect the larger system of structural racism.^{4 5} Political mobilisation might be one effective strategy to make policing fair for all racial groups,³⁸ but antiracist policy achievements are fragile and their continuation is subject to the current political climate.³⁹ For example, several of the national reforms implemented during the Obama administration have been challenged under the Trump administration. Among these, the Trump administration has significantly scaled back the Justice Department’s Collaborative Reform programme, in which local police departments voluntarily collaborated with the Justice Department to improve trust between local police and the public.⁴⁰

In 2020, the law enforcement-related deaths of George Floyd and Breonna Taylor again brought the structural racism in the criminal justice system into the larger political awareness. Black communities may experience increased COVID rates if the current circumstances around the deaths of George Floyd, Breanna Taylor and others lead to lower healthcare-seeking behaviour. This study establishes the need for further investigation into the health effects of law enforcement-related deaths. Further research is needed to better understand the mechanisms involved and to describe cross-institutional impacts of these events, locally and in other jurisdictions. It is important that we continue to acknowledge the breadth of incidents that contribute to collective trauma for communities of colour. This will help guide and support long-term effective interventions that can potentially prevent and address the impact of law enforcement-related deaths on black communities.