Mental illness, impairment, and crisis state in fatal edged-weapon encounters, 2015–2024

Takeaway

Over the decade from 2015 to 2024, at least one of three crisis indicators — mental illness, substance impairment, or suicidal intent — was present in 72.5 percent of fatal edged-weapon encounters. The three indicators frequently co-occur. Despite this concentration of crisis presentations, a Crisis Intervention Team officer was deployed in fewer than three percent of cases where the decedent was coded as suicidal, and in fewer than five percent of calls dispatched as mental health, suicidal, or wellness checks.

Background

The intersection of mental illness, substance impairment, and suicidal crisis with fatal police encounters has been documented across several datasets. Edged-weapon incidents present a particular challenge: the weapon is close-range by nature, encounters often unfold quickly (see EWRL Report 1), and subjects in acute psychiatric or suicidal crisis may not respond to verbal de-escalation in predictable ways.

Crisis Intervention Team (CIT) training — modelled on the Memphis model — has become the most widely adopted programmatic response in North American law enforcement to encounters involving mental health crisis. The present report describes the prevalence and co-occurrence of crisis indicators across the dataset and examines the rate at which CIT and crisis negotiator resources were deployed when those indicators were present.

Data and methods

Dataset. The analysis draws on the full Edged-Weapon Response Lab dataset (N = 1,970), covering fatal police encounters involving an edged weapon documented in the Washington Post Fatal Force database (Washington Post, 2025) and Mapping Police Violence (Mapping Police Violence, 2026) for January 2015 through December 2024.

Mental illness flag (D_Illness). Set when a mental health condition — diagnosed or apparent — was documented in official records, news accounts, or family statements. The flag does not specify a diagnosis or distinguish acute from chronic presentations. Present in 1,204 cases (61.1%).

Impairment flag (D_Impaired). Set when substance intoxication or impairment was documented at the time of the encounter. Present in 601 cases (30.5%).

Suicidal flag (D_Suicidal). Set when suicidal intent was explicitly documented — through statements, prior threats, or formal assessment — in the immediate period preceding or during the encounter. This variable captures what is commonly termed “suicide by cop” when coupled with a fatal outcome, but the coding requirement is documented intent, not an inference from behaviour alone. Present in 676 cases (34.3%). A separate self-harm flag (D_SelfHarm) captures non-suicidal self-injurious behaviour and is present in 204 cases (10.3%).

CIT and negotiator deployment (I_CIT, I_Negotiator). I_CIT is set when at least one officer with Crisis Intervention Team certification was documented as responding. I_Negotiator is set when a trained crisis negotiator was deployed. These flags capture documented deployment and may undercount cases where a CIT-trained officer responded but was not identified as such in source materials.

Suppression. All counts are drawn from the suppressed public aggregate. Cells with fewer than five records are not displayed. For more information see EWRL Methods — Suppression and privacy.

Findings

Prevalence and co-occurrence of crisis indicators

At least one of the three crisis indicators — illness, impairment, or suicidal intent — was present in 1,428 of 1,970 cases (72.5%). Only 542 cases (27.5%) carried none of the three.

The three indicators co-occur frequently. Among illness-coded cases, impairment was additionally present in 440 (36.5%) and suicidal intent in 573 (47.6%). The combination of all three — illness, impairment, and suicidal intent simultaneously — was present in 225 cases (11.4%). The largest single group was mental illness without impairment and without suicidal intent (416 cases, 21.1%), followed by mental illness with suicidal intent and without impairment (348 cases, 17.7%).

Figure 1. Co-occurrence of mental illness, impairment, and suicidal intent flags in fatal edged-weapon encounters, 2015–2024 (N = 1,970).

↓ Figure 1 appears below — jump to chart and data table

CIT and crisis negotiator deployment

Crisis intervention resources were rarely deployed even in cases where crisis indicators were present. Of 676 cases with the suicidal flag, a CIT officer was documented in 19 (2.8%) and a crisis negotiator in a partially overlapping set. Across all 1,970 cases, CIT was deployed in 44 cases (2.2%) and a negotiator in 86 cases (4.4%).

Among calls dispatched in the most clearly crisis-relevant categories — mental health and crisis (42 calls), suicidal and self-harm (125 calls), and welfare and wellness checks (61 calls), totalling 228 dispatched calls — CIT was documented in only 10 cases (4.4%).

Figure 2. CIT deployment in fatal edged-weapon encounters by suicidal flag, 2015–2024.

↓ Figure 2 appears below — jump to chart and data table

Encounter duration and crisis indicators

Encounter duration differs meaningfully between cases with and without mental illness. Among illness-coded cases with a known duration (n = 1,058), 67.0 percent resolved in the S band (under 90 seconds), compared with 74.8 percent of cases without mental illness coding (n = 580). Extended encounters — 90 minutes or more — were present in 6.0 percent of illness-coded cases and 5.2 percent of cases without illness coding.

The suicidal flag is associated with a somewhat longer duration profile. Encounters with suicidal coding show a lower S-band rate (397 of 614 coded cases, 64.7%) than encounters without suicidal coding (748 of 1,026 coded cases, 72.9%), and a higher rate of standoff or barricade involvement: 13.6 percent of suicidal encounters (92 of 676) involved a standoff or barricade element, compared with 10.9 percent of non-suicidal encounters (141 of 1,294).

Figure 3. Encounter duration in fatal edged-weapon encounters by suicidal flag, 2015–2024.

↓ Figure 3 appears below — jump to chart and data table

Weapon type and crisis indicators

The distribution of weapon types differs by illness status. Among illness-coded decedents, the kitchen knife is the most common weapon (351 of 1,204 illness cases, 29.2%), compared with 17.1 percent of non-illness cases (131 of 766). The generic knife category leads among non-illness cases (301 of 766, 39.3%) but accounts for a lower share of illness-coded cases (238 of 1,204, 19.8%). The kitchen knife finding is consistent with the domestic and household settings associated with many crisis presentations.

Interpretation

What the findings show. The overwhelming majority of fatal edged-weapon encounters involve a decedent with at least one documented crisis indicator. Mental illness, impairment, and suicidal intent frequently co-occur, and the three-way combination is not rare (11.4% of all cases). Despite this concentration, CIT deployment in suicidal encounters — and in calls dispatched as crisis calls — is in the low single digits. The gap between crisis indicator prevalence and crisis resource deployment is substantial.

What the findings do not show. This dataset does not capture the proportion of crisis calls that were recognised as such at dispatch, nor does it identify cases where CIT-certified officers were present but not documented as such. The CIT deployment figures are therefore a lower bound on the frequency of CIT involvement and an upper bound on the size of the gap; the true gap may be somewhat smaller than the raw figures suggest. The dataset also does not record whether CIT training or deployment changed the course of any encounter.

What would change the conclusion. Data on CIT deployment in the broader population of edged-weapon calls — including non-fatal resolutions — would be needed to assess whether CIT deployment is associated with different outcomes. Administrative records documenting CIT certification and dispatch protocols by agency would substantially improve the precision of deployment estimates.

Limitations

The illness, impairment, and suicidal flags are derived from publicly available records and may reflect reporting and salience biases. Conditions not mentioned in source documents are not coded, and some proportion of the 27.5 percent with no flags almost certainly had unrecorded mental health histories. The coding threshold for D_Suicidal — explicit documentation of intent — is conservative; encounters with behavioural indicators of suicidal crisis but no explicit documentation are not captured.

CIT and negotiator deployment (I_CIT, I_Negotiator) depend on documentation in the same public sources and are likely undercounted. The absence of a CIT record in source materials does not confirm that no CIT-trained officer was present.

How to cite

Machacynski, B. (2026). Mental illness, impairment, and crisis state in fatal edged-weapon encounters, 2015–2024 (EWRL Report No. 3, Version 1.0). Edged-Weapon Response Lab. https://droptheknife.org/reports/mental-illness-impairment-and-crisis

Data snapshot. S-20261002, deposited at Zenodo https://doi.org/10.5281/zenodo.22650264.

References

Mapping Police Violence. (2026). Mapping Police Violence database [Data set]. Retrieved March 4, 2026, from https://mappingpoliceviolence.org

Washington Post. (2025). Fatal Force database [Data set]. Retrieved August 11, 2025, from https://github.com/washingtonpost/data-police-shootings

How to cite

Edged-Weapon Response Lab. (2026). Mental illness, impairment, and crisis state in fatal edged-weapon encounters, 2015–2024. In EWRL Research Reports (Version 1.0). https://droptheknife.org/reports/mental-illness-impairment-and-crisis