Procedural Deficiencies in Cause of Death Reporting Systems Cause Extensive Underreporting of Fatal Police Violence

by Casey J. Bastian

Our nation is experiencing a recent period of forced enlightenment. The scope of violent or fatal encounters with American law enforcement is now considered by many experts to qualify as a public health crisis. A fact that many people have been forced to recognize in the wake of multiple, high-profile police killings over the last several years. The reality of police brutality in the Land of the Free isn’t a nascent phenomenon, but it has been effectively denied for decades. Specifically, members of our marginalized communities have experienced the worst of this reality while that violence was overlooked. Very rarely does the responsibility for this atrocious behavior land where it is warranted: on the bloody hands of abusive, and often unaccountable, law enforcement.

An analysis of the latest data from 2019, published in an October 2021 report by The Lancet, demonstrates that the U.S. accounts for an estimated 13.2%, or 1,150, of all the global deaths arising from police conflict. However, America only represents a mere 4% of the global population. Police conflict accounted for more deaths of American males in 2019 than: environmental exposure (931 deaths), Hodgkin Lymphoma (835), testicular cancer (486), and sexually transmitted diseases (37), each a public health concern. Violence is now being considered a public health issue as it affects mental and physical health while undermining community safety and individual well-being. That violence is most detrimental when it comes at the hands of those tasked with ensuring our safety and well-being. Worse, violence, particularly police-initiated violence, is preventable. This is why a large number of advocacy groups, community organizations, and governmental agencies condemn police violence and identify its seeming endemic racism as nothing less than an urgent public health crisis. These organizations include the American Public Health Association, the American College of Physicians, the American Medical Association, and the Institute for Health Metrics and Evaluation.

The reality is that the level of fatal police violence may be worse than we believe. The Lancet study found that 55% of killings attributable to police violence between 1980-2018 are likely misclassified and underreported in the government-run reporting system. Researchers compared statistic from the U.S. government’s National Vital Statistics System (“NVSS”) database to those databases compiled by three publicly-run groups. Researchers found that between 1980-2018, there were 30,800 deaths in America caused by police; 17,100 more than was reported during that period in the NVSS. The researchers noted that “the misclassification of police violence in NVSS data is extensive.”

The data analysis published in The Lancet was conducted by researchers from the Global Burden of Diseases, Injurious, and Risk Factors (“GBD”), titled, “Fatal police violence by race and state in the USA, 1980-2019: a network-meta regression” (“Study”). The Study focused on three primary objectives. First, the data were examined to reveal the presence and extent of misclassification and underreporting of deaths resulting from police violence in the NVSS. Second, revised estimates of such deaths in America culled from the examined independent databases were reported. Finally, a method is proposed to correct the data deficiencies within the NVSS. The GBD defines “police conflict,” used to ascertain the 2019 numbers as deaths including “civilians killed by police, police killed by civilians, and government run executions.” The Study primarily focused on “police violence” defined as “police-related altercations leading to death or bodily harm.” However, the Study focused and reported on only the statistical data resulting from police violence-related deaths within the examined data. 

Most high-income nations like the U.S. rely on vital registration systems run by the government to collect cause of death data. Many similar systems are considered highly reliable sources of death data despite the obvious source of potential conflict. The data can appear to be susceptible to misrepresentation when the government responsible for the actual police violence is the same entity responsible for accurately reporting that same violence. The Study reveals that the national compilation of the state death data within the NVSS has been misclassified, resulting in the observed underreporting. This particularly diminishes the full impact of American police violence on our Black, Hispanic, and Indigenous communities.

The NVSS is deficient in its reporting because of the classification procedures not requiring the definitive information that a cause of death may specifically be the result of police violence as defined and analyzed by the Study; it is not due to any systemic racism in the NVSS itself. This is revealed by the Study and is made apparent by the proposed improvements in reporting methodology. However, we can never reduce or prevent police violence against any citizen if correct data are not available for formulating necessary measures to do so. And when the numbers reveal that minorities are most impacted by such violence, a lack of accurate data prevents exposure of possible endemic racism where it may lie—in the law enforcement apparatuses of America. 

GBD researchers began by locating databases that might capture fatal police violence data more comprehensively than the NVSS. On June 2, 2020, researchers used Google and Google Scholar to search for the terms “police violence OR killing OR shooting OR conflict databases.” Any database referenced in the search results was required to meet the following criteria prior to its data being analyzed: “(1) inclusion of both firearm and non-firearm-related deaths, (2) including of state and race or ethnicity, and (3) improve coverage compared with the NVSS, if the database also met the first two criteria.” This research strategy is known as “open-source methodology” and allows for capturing police violence death data using publicly available information. The U.S. Bureau of Justice Statistics considers public databases to be more reliable than government databases due to official under-reporting, making such methodology preferable when gathering data to be analyzed. The three databases fitting GBD criteria were Fatal Encounters, Mapping Police Violence, and The Counted. The GBD chose to forego databases from The Washington Post’s Fatal Force, The National Violent Deaths Reporting System, and the Arrest-Related Deaths Program. The three selected databases include data from all 50 states and Washington, D.C.; none of them contains information from U.S. territorial possessions.

Fatal Encounters is a non-profit organization that collates news reports and public records requests of people killed during encounters with police between 2005-2019. It also contains data from 2000–2004, but this data were not considered in the studies due to questions of data completeness; namely, racial and ethnic information that was not available. Mapping Police Violence is non-profit research collaborative that mirrors Fatal Encounters but has only collated data from 2013-2019. The Counted is a project of The Guardian, collating similar data from all law enforcement agencies for the years 2015-2016.

The NVSS is a government-run system coordinated by the National Center for Health Statistics that uses death certificates listing “legal intervention” International Classification of Diseases (“ICD”) codes for underlying causes of death identified by medical examiner or coroner determined cause of death. The NVSS encompasses 39 years of data. The use of ICD codes by those certifying death certificate information, which is done in accordance with the World Health Organization’s published code selection rules, is a standardization meant to ensure classification uniformity in cause of death reporting systems. These rules require that a death due to police violence be classified under these legal intervention codes. These codes are defined as “injuries inflicted by the police or other law enforcement agents, including military on duty, in the course of arresting or attempting to arrest lawbreakers, suppressing disturbances, maintaining order, and other legal action.”

A physician completes a death certificate under normal circumstances, but in cases of police violence, it is to be a coroner or medical examiner. When filling out a death certificate, one certificate text field is critically important: “Describe how the injury occurred.” If no information is given that the decedent was killed by the police, the death does not receive a legal intervention code. This results in the underreporting by misclassification found in the NVSS as evidenced by the Study’s analysis. Researchers found that this specific text field “is open-ended and comes with no explicit instructions to mention police involvement, and a certifier might lack the knowledge or training to fill out the form correctly.” Underreporting is not only just a result of a lack of professional training causing misclassification.

The National Association of Medical Examiners conducted a survey of its members in 2011 and found that 22% reported that they had “been pressured” by government officials “to change the cause or manner of death on a certificate.” If this is done, a death from police violence is classified as a typical homicide, which also leads to underreporting of police violence. This is also a result of certifiers being embedded within the police departments as well—a substantial source of potential conflicts of interest. The Study did not just reveal deficiencies in any one system, it exposed several significant data limitations within the three publicly available databases that were examined as well.

The three databases cover a collective 20-year period from 2000-2019. Significantly less than the NVSS database. This limits the scope of accurately estimating police violence within the open-source databases. Fatal Encounters is the longest-running of the three databases but suffers from an overbroad case definition that includes all deaths arising from police violence encounters, with no requirement of police culpability. This makes the data accurate for recording the fact of any death occurring during the encounter but does not assist in determining deaths resulting from police conduct itself, e.g.—if it was the officer who died, that would be a death occurring during the encounter but does not assist in extrapolating the rate of civilian deaths as a result of police violence.

In the alternative, Mapping Police Violence and The Counted only consider civilians killed and include a specific definition of police violence. But these two databases specifically only encompass data over a collective seven years between 2013-2019. As mentioned, the Fatal Encounters data from 2000-2004 had to be wholly disregarded. GBD researchers were not just focused on revealing the inaccuracy of data reporting within a given system but also disclosing the true disparity of police violence as experienced by minorities. These analysis goals revealed further limitations in each database as well. Race or ethnicity data were identified as missing in each: Fatal Encounters is missing 22% of such data, the NVSS (15% missing), Mapping Police Violence (9% missing), and The Counted (2% missing of the 2,239 deaths in that database).

The GBD researchers identify The Counted as “The Gold Standard” in public database compilation because of its “open-source methodology, case definitions of police violence, and high completeness on race and ethnicity,” even though the data span only the two-year period of 2015 and 2016. Researchers extracted and standardized “age, sex, US state, year of death, and race and ethnicity of each decedent for all data sources.” The variation of raw data reported for race and ethnicity exists across data sources, U.S. state, and time, presented a substantial challenge for standardization. Researchers chose four categories for standardization: non-Hispanic Black, Non-Hispanic White, non-Hispanic of other races, and Hispanic of any race (Indigenous peoples fall into this category).

For extracting data, the Study used network-meta regression (“NMR”) to “quantify the biases of the three other datasets compared with The Counted.” Put simply, NMR is a “mixed effects log-transformed linear regression” that inputs “all input pair wise comparisons between datasets to deduce the relative biases between them,” increasing the amount of data available by “5.5 times.” The meta-regressed data revealed several statistics, some quite disturbing. That there appears to be a substantial, disparate likelihood of police violence experienced by minorities, particularly non-Hispanic Blacks, is indisputable. 

Some notable statistics found include: men account for almost 97% of all police deaths. Age-standardized mortality rates (“rates”) are highest for non-Hispanic Blacks at .69 per 100,000 persons, followed by Hispanic of any race (.35 per), non-Hispanic White (.20 per), and non-Hispanic of any race (.15 per). When aggregating all races, the rate increased from .25 per 100,000 in the 1980s to .34 per in the 2010s; a 38.4% increase in overall police deaths.

The states with the highest rates were Oklahoma at 1.22 per 100,000, followed by Washington, D.C. (.70 per), Arizona (.70 per), Alaska (.60 per), Nevada (.57 per), and Wyoming (.56 per). For every decade, non-Hispanic Blacks had the highest rates and was especially prevalent in Oklahoma, Alaska, West Virginia, Utah, and D.C.

The underreporting of such data distorts and diminishes the greater public health issue that is fatal police violence with associated disparities in race and ethnicity. The estimated rates of underreporting in the NVSS for incidents of fatal police violence is 56.1% for non-Hispanic Blacks, 56.1% for non-Hispanic Whites, 50% for Hispanic people of any race, and 32.6% for non-Hispanic of any race. The top five states with the highest rates of underreporting numbers of deaths were Oklahoma underreporting 83.7% of all such deaths misclassified, followed by Wyoming (79.1%), Alabama (76.9%), Louisiana (75.7%), and Nebraska (72.9%).

Identifying and correcting the current problems with underreporting due to misclassification across all reporting systems is the first step in imposing accountability and transparency in policing. Increasing the use of open-source data-collection initiatives will provide the opportunity for more research on these issues. Providing this information to policy makers will allow for updated policies and procedures to alleviate police violence overall. Allowing for the creation of targeted and meaningful changes to public safety and policing that in turn will assist in diminishing the disparate police violence that impacts minority communities.

Some basic changes must first be made in the classification procedures of all reporting systems. This includes improved training and more precise instructions on how to document police violence in the text fields of death certificates. To avoid pressure by law enforcement agencies being placed on coroners or medical examiners, they must work independently from law enforcement. This will eliminate one source of incorrect assignments of cause of death. In addition, the GBD researchers suggest that forensic pathologists must investigate and testify in cases where police violence is a cause of death, and these professionals should be awarded whistleblower status under the law to ensure reporting of pressure and undue influence from government agencies.

Improving the issue of underreporting due to misclassification is one important area that has been clearly identified by many studies similar to this one undertaken by the GBD. But it will also require that the field of public health turn its attention to fully eliminating the burden of police violence in America. Solutions found through evidence-based research and advocacy work are needed. It may be necessary that we continue to rethink policing in this country. Our law enforcement is all too often militarized. Other countries around the world have recognized this trend, and 19 nations, including Norway and the United Kingdom, eliminated armed law enforcement. It has not resulted in mass chaos. Many of our country’s leaders suggest that it is unthinkable to defund, disarm, or abolish militarized police; that attitude does not reflect reality.

The difference in police violence experienced in such countries is staggering. Not one person died from police violence in Norway in 2019, and only three died in the United Kingdom and Wales between 2018-2019. It is vital that accurate statistics are available to provide the evidence needed to make this a reality in America as well.  

Sources: thelancet.com, reason.com

Subscribe today

Already a subscriber? Login