Health impacts of chemical irritants used for crowd control: a systematic review of the injuries and deaths caused by tear gas and pepper spray

Search strategy

We searched PubMed, Toxnet, JSTOR, and Scopus using search terms cross-referenced with the MeSH database with no language restrictions [30]. We expanded our search to include non peer-reviewed publications as well as relevant reports identified by experts in the field. Gray literature searches were also conducted using reference lists of relevant articles and recommendations from experts. We included data from all types of studies, including experimental and observational studies and case series with at least five subjects.

Study selection

Articles were included if they documented injuries, deaths, or other medical or psychological health consequences of chemical irritants on human subjects and were published between January 1, 1990 and March 30, 2015. We included studies of cohorts of all ages, genders, and ethnicities. We included data from all contexts of chemical irritant use, including demonstrations and protests, riots, sporting events, prisons, arrests, and accidental exposures, as well as military or police training events. We excluded studies that lacked adequate documentation on injuries, were not accessible for full text review, or were animal and cadaver studies.

Titles and abstracts of the articles were screened for relevance. Full texts of all potentially relevant articles were reviewed against our inclusion criteria.

Data extraction

Data from all eligible articles were then extracted and compiled in a database (Microsoft Excel for Mac 2011 v14.4.1). All articles were read and coded by two authors (RH and MD). Disagreements were resolved by discussion between the authors. For each study, we identified the chemical agent, deployment mechanism (spray versus aerosol or versus mechanical injuries from the projectile munition), region/country, demographic characteristics, and study setting. We categorized the outcome for subjects as recovered, permanently disabled, or dead, and classified each injury by severity and body system. Injury severity was coded based on the acuity and the resources required to manage that injury. Minor injuries were transient symptoms that may not be present on physical exam or are expected side effects of chemical irritants, such as blepharospasm, lacrimation, mild respiratory distress, sore throat, or nausea. Moderate injuries were those that were unexpected from previous published data on chemical irritants, were evident on physical exam, or lasted longer than expected, but may not require management by a health professional. Injuries such as persistent skin rashes or erythema, first-degree burns, conjunctivitis or eye injuries, oropharyngeal edema, persistent respiratory symptoms, and vomiting were classified as moderate injuries. We classified as severe injuries those that necessitate professional medical care, such as lacerations requiring sutures, second- or third-degree burns, airway obstruction, severe ocular trauma, cardiopulmonary disease, or abdominal injuries requiring medical or surgical management. Injury data was only included if it was documented by a medical professional. Injuries that were reported by patients, without any documentation, were excluded.

We utilized the NIH Quality Assessment Tool to classify each article as poor, moderate, or high quality [31]. This tool was chosen to standardize the quality of case series and observational studies, which made up a majority of the identified articles.

Data analysis

We conducted descriptive analysis of injuries from chemical irritants to categorize the range of injuries and their severity. We also evaluated mediating or moderating environmental or practical factors that may have increased or reduced injuries. We expected significant heterogeneity and quality limitations that would preclude pooled data analysis or a direct comparison of the different chemical irritants.

Demographic analysis

Of the total of 31 studies included in the analysis, 16 were retrospective cohort studies, seven were prospective cohort studies, six were case series and two were non-peer reviewed reports from reputable human rights organizations. The number of subjects ranged from two to 3697 (median 31) (one study met the inclusion criteria of five subjects, and although several of the subjects sustained injuries from another weapon, the study was nonetheless included). In studies in which gender was reported, 57% of subjects were male and 43% were female. In studies in which age was reported, the age ranged from 3 months to 94 years, with an mean age of 25.7 years. The injury context included protests (10), arrests (five), military or police training exercises (five), accidental exposures (five), and a detention center riot (one); some of the studies included injuries in more than one context. The eligible studies included data from 11 countries and were published between 1993 and 2015 (Fig. 2).

Twenty-six studies included data on injuries caused by Agent CS and 14 included data on injuries caused by Agent OC. Sixteen studies evaluated dry aerosolized forms, such as grenade-type deployment of the chemical agent and 15 included sprays formulated with solvents. Seven of the studies recorded injuries that resulted from the projectile munition containing chemical irritant causing direct trauma to subjects.

Analysis of injuries and deaths

In the included articles, a total of 5910 people were exposed to chemical irritants and sought medical attention, of whom 5131 (87%) suffered injuries or died as a result of the exposure. Of those who suffered injuries, two people died and 67 (1.3%) suffered permanent disability. The majority fully recovered from their injuries (98.7%).

Deaths

Two deaths were documented in the selected articles. A report from Bahrain documented the case of a man who died of respiratory arrest after agent CS aerosol was fired inside his home. In another case, the chemical irritant projectile munition contributed to one death from traumatic brain injury after protests in Nepal. There were no deaths associated with agent OC.

Permanent injuries and disabilities

Fifty-eight people experienced permanent disability (Fig. 3). Eighteen of the disabilities were secondary to traumatic injuries from the projectile munitions. These included globe ruptures and blindness (four people), traumatic brain injury resulting in a persistent vegetative state (one person), limb amputations (three people), and functional loss of limbs (10 people). Persistent psychiatric symptoms were documented in 14 people and persistent symptoms of asthma and other respiratory complaints were reported in 23 people. Chronic dermatological conditions such as hypersensitivity reactions were documented by skin testing in three people. In one study of 297 individuals seeking care and/or evaluation of injuries following the 2013 Gezi Park protests in Turkey, 117 psychiatric evaluations were conducted. Of those, 50 (43%) met diagnostic criteria for acute stress disorder, 27 (23%) met diagnostic criteria for post-traumatic stress disorder (PTSD), and nine (8 %) met diagnostic criteria for major depressive disorder [62].

Injuries

There were 9261 documented injuries, with multiple injuries occurring in each individual. In total, 6878 (74.2%) of the injuries were categorized as mild, 1582 (17%) were moderate injuries, and 865 (8.7%) were severe injuries (Fig. 4). While many body systems were affected, the majority of injuries were to the skin, eyes, and cardiopulmonary system. Injury severity varied depending on body system affected (Fig. 5). Severity followed a pattern of fewer injuries among the higher severity categories for dermal, ocular, neurological, cardiopulmonary, and intra-abdominal injuries. For instance, while the majority of skin injuries (2539) were mild, 351 of the injuries were moderate and 180 were severe. Similarly, for cardiopulmonary injuries, there were 1220 mild injuries, 328 moderate injuries and 131 severe injuries. On the other hand, there were more severe neurological injuries (12%) than moderate neurologic injuries (1 %). Injuries to the musculoskeletal system, as well as psychological injuries, were all categorized as severe, based on the definition employed.

Chemical agent

The nature of chemical irritant exposure and injuries is also related to the chemical agent used, mechanism of deployment, environmental conditions, and context of use. Fourteen studies dealt exclusively with agent CS and 10 studies exclusively studied agent OC. Three studies included injury data from both chemical agents or did not differentiate between the two. Four studies reported exclusively on traumatic injuries from the projectile munition, while three other studies reported some injuries from the projectile munition among other injuries from the chemical agents themselves. Among 7156 documented injuries specifically from agent OC, only 6 % were categorized as severe. In contrast, 27.9% of 1148 injuries from agent CS were categorized as severe (Fig. 6).

Other factors that may impact injury severity

To assess for other factors that may impact injury severity, we utilized a qualitative approach by perusing the articles for data that may not fit into categorical variables but would regardless be relevant. Several of these factors were noted in the article text and are highlighted.

Deployment mechanism

The included studies documented injuries secondary to both dry aerosolized and soluble spray forms of both chemicals. Proximity to the area where the chemical was released and the force of the propellant affected outcomes [38, 40, 47, 53, 60]. We identified 5366 mild injuries, 884 moderate injuries, and 483 severe injuries from spray forms of agent CS and OC. Among the injuries from spray forms of chemical irritants, 7.2% were severe. We documented 1512 mild injuries, 676 moderate injuries, and 281 severe injuries from aerosolized forms of agent CS and agent OC. Among injuries from aerosolized forms, 11.6% were severe. Comparative analysis of the deployment mechanisms using pooled data was not conducted, given the concern for confounding factors.

The use of projectile munitions was documented to cause 231 injuries, of which 63 (27%) were severe. There were 73 traumatic injuries to the head and neck, including at least four people who lost vision in an eye due to projectile munition trauma. We documented 45 injuries to the torso (chest, abdomen, back, and genitalia). There were 61 upper extremity injuries and 34 lower extremity injuries (including at least three people requiring amputations and 10 with severe functional loss of a limb due to neurovascular injuries). Eighteen dermal injuries (8 %) included bruises, lacerations, and heat burns.

Other factors

Several other factors were documented as exacerbating the potential for injury, but they lacked detailed data for analysis, such as documentation of specific injuries. Utilizing the weapons in confined spaces and in areas where people could not easily escape potentially increased the exposure to the irritant either in quantity or over time [54]. One study in a detention center suggested that the excessive number of injuries may have been caused by the crowded and enclosed setting, which offered no opportunity for people to escape [33]. Use of chemical irritants in areas with high heat or humidity potentially exacerbated skin irritation, and windy conditions risked the contamination of law enforcement officers, bystanders, or nearby residences and businesses [33, 43, 53]. One study noted that the use of agent CS for military training on a particularly humid day, followed by strenuous exercise by trainees, may have caused severe respiratory injuries which resulted in several people requiring ICU-level care [53]. Direct targeting of the face and eyes by hand-held spray has been noted to cause trauma and toxicity to the cornea and conjunctiva of the eye [40, 41, 45, 50].

Policy implications and recommendations

The legal protections of the rights to freedom of expression and peaceful assembly, along with general principles on the proportionate use of force by law enforcement, provide some general guidance that the use of chemical irritants, along with other crowd-control weapons, should be limited. Specifically, CCWs should only be used in situations where particular individuals pose an imminent violent threat, or where a protest requires dispersal because of widespread violent acts that pose an imminent threat to public safety [74, 75]. In most situations where we found these weapons being used, neither of these conditions was documented. The use of chemical irritants as crowd-control weapons must be considered in the broader context of human rights, public safety, use of force, and law enforcement practices necessary to maintain order in the context of demonstrations. Open communication with demonstrators and the community, arrests of violent individuals, and safeguards for legal demonstration may obviate much of the demand for these chemicals. Given the frequency of serious injury, disability, and death, the use of chemical irritants should be strictly limited to situations of imminent harm that cannot be policed effectively with safer methods.

The Chemical Weapons Convention (effective since 1997 with 192 state signatories excluding only Egypt, Israel, North Korea and South Sudan) prohibits the use of these riot control agents’ during warfare. Significant questions exist on the legality of military use of these weapons in civilian protest. We recommend that law enforcement and the military be obligated to maintain transparent and accurate data on use-of-force incidents, particularly those that employ chemical irritants and other crowd-control weapons. Active surveillance of injuries caused by chemical irritants is vital for manufacturers, law enforcement, and the community in order to understand the risks and dangers of these weapons. This data should be available and accessible to the public for independent analysis.

We also note that combinations of OC and CS are becoming more common, both in spray and aerosol forms as well as within projectiles such as the “pepper ball” [16, 76, 77]. Several newer agents are also in development, including agents CS1 and CS2 (which may extend the half-life of CS or facilitate higher dermal penetration) and agent CX, which is reported to be more potent than agent CS [16]. Each of these potential enhancements to weapons may compound the already large number of injuries. It is important that we address the human costs of current chemical irritants before developing new, more potent ones.

While making law enforcement protocols publicly available may not be possible due to security risks, we recommend that police and military departments make every effort to communicate with health workers and the community in order to minimize potential injuries from chemical irritants and to maintain trust. Training of police officers must include education on human rights principles and the obligation of the police to protect peaceful protestors. Police should also be trained in the dangers of chemical irritants, guidelines on the safe utilization of chemical irritants, the risks of repeated exposures, environmental factors, and the risks of direct trauma from poorly-aimed projectile munitions, as well as other risk factors.

Limitations

Our systematic review had several limitations. In the absence of systematic reporting requirements on deaths and injuries in crowd-control settings, it is likely that we have largely underestimated the prevalence of deaths and injuries. The limited follow-up in many of the articles also highlighted the lack of data on the chronic health impacts of these weapons, which are likely underreported. We note that in our attempt to ensure valid and reliable injury data, we have not accounted for a large number of injuries and deaths from chemical irritants that have been reported by the news media, by social justice organization reports, and by social media, many of which include photographic or videographic evidence of injuries. We also excluded reports that lacked injury specifics or clear causation from chemical irritants, but were likely linked to chemical irritant utilization. We also excluded the significant number of case reports in the published literature to avoid biasing our results towards the publication of the most severe injuries. In addition, there is wide variability in how weapons are used and in the specific concentrations of chemical agents; there is also a lack of data on the number of people exposed. Given these concerns, we were not able to calculate population estimates of the impact of chemical irritants or compare any specific agents or numerical study results. However, severe injuries from chemical irritants are not rare or isolated incidents. They have occurred in many nations and under different types of regimes and law enforcement protocols.

This review was also limited by the quality and methodology of the available literature on chemical irritant injuries. There are several potential biases, including the potential over-publication of the most dramatic incidents and independent limitations on individuals’ decision to seek medical care. On the other hand, difficulty in gathering and publishing data in repressive regimes may limit the availability of injury data from many instances of chemical irritant utilization. There was also significant heterogeneity in the participants and medical treatments in different regions and clinical settings. There was considerable methodological variability in the study designs and settings. Each setting had variable standards on the use of chemical irritants. However, the overall quality of the studies was comparable to observational and case series-type studies addressing chemical irritants. Given the multiple confounding factors, we could not compare the chemical agents or deployment mechanisms. Included studies did not provide enough data to reliably estimate the risk of injury from any given chemical irritant in an exposed population.