How and Why FLAME Was Built
The research base, theoretical frameworks, design decisions, and scoring methodology behind the FLAME program.
What Has — and Has Not — Been Established
FLAME is built on a strong evidence base for lethal means restriction as a suicide-prevention strategy (summarized below). That evidence supports the program's core premise: reducing firearm access during a crisis saves lives. It does not, by itself, establish that the specific FLAME score is a reliable or predictive measurement. Those are separate empirical questions that require dedicated study.
What is established
- The scientific case for reducing access to lethal means during periods of elevated risk (Lubin et al., 2010; Barber & Miller, 2014).
- The acceptability and effect of brief, non-judgmental lethal means counseling on storage behavior (Runyan et al., 2016).
- The theoretical rationale for treating firearm accessibility as a multiplier of other risk factors (the Integrated Motivational-Volitional model).
What is not yet established (and is the subject of planned research)
- The point values and multipliers. The risk-factor weights (for example, suicidal ideation = 12, dementia = 7), the accessibility multipliers (0.07 through 1.0), and the tier cut-offs reflect clinical and scientific judgment by the developer. They have not been empirically derived or weighted from outcome data.
- Reliability. Because FLAME is intended to prompt reflection and behavior change, conventional test–retest reliability is not appropriate; planned work uses short-interval retest before any feedback, internal-consistency estimation (McDonald's omega), and a no-feedback control condition to separate measurement error from intervention effect.
- Validity. Content validity (expert/Delphi review with I-CVI and S-CVI benchmarks), known-groups and concurrent validity, and — as a dedicated capstone requiring future funding — predictive validity against behavioral and clinical outcomes.
The calibration tables on this page confirm that the algorithm computes the scores it was designed to produce across its range. That is an internal consistency check of the code — not evidence that the scores predict real-world outcomes. They are labeled accordingly below.
Background: The Firearm Suicide Crisis
Firearms are used in approximately 54% of all U.S. suicide deaths — the numerically dominant method of suicide mortality. The case fatality rate for firearm suicide attempts is approximately 85%, compared to roughly 5% for the most common non-firearm methods. This lethality difference means that firearm access during a suicidal crisis is a critical variable — not because suicidal crises always lead to death, but because they so often do when a firearm is immediately accessible.
Veterans face disproportionate risk: approximately 70% of veteran suicides involve firearms, and veterans are about 1.5× more likely than age-matched civilians to die by suicide. In Wisconsin, firearm suicide accounts for a significant share of total suicide mortality, with rural populations, agricultural communities, and veterans at elevated risk.
The Science of Lethal Means Restriction
The most empirically supported suicide prevention strategy is reducing access to lethal means during periods of crisis. Suicidal crises are typically time-limited and impulsive — most people who survive do not attempt again. A landmark study of Israeli military policy found that requiring soldiers to store firearms off-base during weekend leave reduced firearm suicide rates by approximately 40%, with no compensatory increase in other methods. (Lubin et al., 2010) Individual-level lethal means counseling (LMC) — specifically the CALM curriculum — has demonstrated that brief, non-judgmental conversations about storage significantly increase safer storage practices following clinical encounters. (Runyan et al., 2016)
Theoretical Frameworks
Social-Ecological Model
FLAME is grounded in the social-ecological model of suicide prevention, which addresses risk and protective factors across individual, relationship, community, and societal levels. FLAME operates primarily at the individual and relationship levels — the most amenable to rapid intervention.
Integrated Motivational-Volitional (IMV) Model
The IMV model distinguishes between defeat/entrapment states that motivate suicidal ideation and the volitional factors that enable action. Critically, the IMV model identifies access to means as a primary volitional enabler — the bridge between suicidal intent and completed action. FLAME's multiplier-based scoring architecture directly operationalizes this: firearm accessibility is the mechanism through which all other risk factors gain or lose their lethality.
Transtheoretical Model (Stages of Change)
Individuals exist at different stages of readiness for behavior change. FLAME's readiness and openness questions explicitly assess motivational stage, and the feedback layer is calibrated to the respondent's stated readiness — not assuming uniform motivation across all users. (Miller & Rollnick, 2013)
FLAME v5 Scoring: The Accessibility-Multiplier Model with Configuration Rules
FLAME v5 builds on the multiplier model introduced in v4, adding six mandatory floor and cap rules that enforce clinically meaningful score ranges based on storage configuration, ammunition accessibility, and load status. These rules override the base formula when the physical configuration of firearms in the home represents a defined risk level regardless of other factors.
How the Score Is Calculated
Step 1 — Combined Risk Score (0-65): Three domains are summed additively. These scores represent stable household risk independent of storage configuration.
| Domain | Max Points | Key Variables |
|---|---|---|
| Physical Access | 15 | 10-year-old probe, household composition (children, elderly, disability) |
| Household Risk Factors | 35 | 14 weighted stressors — suicidal ideation (12 pts), domestic conflict (9), dementia (7), suicide bereavement (8), and 10 others |
| Current Crisis Window | 15 | Stress slider, hard-period identification, impulse-access time |
Note on the weights: These point values are face-valid clinical estimates, not empirically weighted coefficients (see Validation Status). Some items warrant particular caution. For example, the developmental-disability item (autism, 5 points) reflects narrow literature on elevated firearm-injury risk under specific conditions; it is not intended to characterize neurodevelopmental conditions as inherently dangerous, and both its wording and weight are flagged for expert review during content-validity work. Weights and item wording may change as validation proceeds.
Step 2 — Accessibility Multiplier (M, 0.0-1.0)
The multiplier represents how accessible firearms are right now, from completely inaccessible (off-site, M=0.0) to maximally accessible (loaded within arm's reach, M=1.0). The multiplier is derived from primary storage method, adjusted by the behavioral anchor (what the respondent actually did last time) and ammunition separation practices. For households with firearms in vehicles or outbuildings, the highest multiplier across all locations is used.
| Storage Configuration | Base M | Notes |
|---|---|---|
| Off-site storage (gun shop, range, trusted person) | 0.0 | No access — score = 0 (floor 5 if risk factors present) |
| Double locked — safe/case + trigger/cable lock | 0.07 | Two independent barriers; lowest home-storage M |
| Locked safe or steel cabinet only | 0.15 | Single container barrier, no additional trigger lock |
| Locked hard case only | 0.28 | Portable container; somewhat lower physical resistance than fixed safe |
| Trigger or cable lock only (no container) | 0.45 | Firearm reachable; lock on firearm itself only |
| Hidden, no lock (drawer, closet, under bed) | 0.75 | No physical barrier — concealment only |
| Accessible, loaded, within arm's reach | 1.0 | Maximum access — no barrier, no steps required |
| Vehicle accessible (no lock) | 0.95 | Highest multiplier wins across all storage locations |
| Vehicle locked / outbuilding locked | 0.55 / 0.40 | External storage — barrier present |
Step 3 — FLAME Score with Configuration Floor/Cap Rules
FLAME Score = (Combined Risk ÷ 65) × 100 × M². This formula produces the base score. Six configuration-specific rules then apply mandatory floors and caps that override the formula when the storage/ammo/loaded state represents a clinically defined risk level regardless of what the formula produces.
| Configuration | Rule | Score Bound | Rationale |
|---|---|---|---|
| Off-site storage, no risk factors | — | = 0 | No means, no score |
| Off-site storage, risk factors present | Floor | ≥ 5 | Risk exists independent of access |
| No lock + ammo not locked + any household risk | Floor | ≥ 65 | Unsecured firearm with ammunition available — directly actionable threat |
| Loaded + no lock (hidden or accessible) | Floor | ≥ 75 | Highest single-configuration risk — no steps required to discharge |
| No lock + ammo locked + unloaded | Cap | ≤ 35 | Two steps required to discharge: reach ammo, load firearm |
| Loaded + trigger/cable lock only | Cap | ≤ 35 | Lock present; highest-resistance "loaded" configuration |
| Loaded + locked case | Cap | ≤ 30 | Container barrier reduces impulsive-access probability |
| Loaded + locked safe | Cap | ≤ 25 | Fixed safe provides stronger resistance than portable case |
| Loaded + double locked | Cap | ≤ 20 | Two independent barriers — meaningful access delay even when loaded |
| Unloaded + container lock + ammo not locked | Cap | ≤ 35 | Good practice; ammo accessible reduces protection level |
| Unloaded + locked case + ammo locked | Cap | ≤ 25 | Three steps required: open case, retrieve ammo, load |
| Unloaded + locked safe + ammo locked | Cap | ≤ 20 | Best single-lock practice |
| Unloaded + double locked + ammo locked | Cap | ≤ 10 | Optimal storage — maximum barriers, unloaded, ammo separate |
Risk Tier Classification
| FLAME Score | Tier | Clinical Meaning | Feedback Approach |
|---|---|---|---|
| 0-20 | Low Risk | Secured storage; low household stressors | Maintenance and monitoring; protective factors noted |
| 21-45 | Moderate Risk | Meaningful barriers present; some stressors | Targeted storage upgrades; specific resource connections |
| 46-64 | Structurally unreachable | See structural gap note above | — |
| 65-74 | High Risk | Rule 1 zone — unlocked firearm, ammunition accessible, any household risk | Concrete actions with projected score changes; Rule 1 floor explanation |
| 75-100 | Critical Risk | Rule 2 zone — loaded firearm with no lock, or maximum combined risk | Crisis resources, trusted-person prompting, Wisconsin Gun Shop Project direct linkage |
Calibration Scenarios (Internal Consistency Checks, 5-Point Increments)
The following calibrated scenarios confirm that the algorithm computes the scores it was designed to produce across its full range. These are internal consistency checks of the code, not empirical validation — they demonstrate that the implementation matches the intended design, not that the resulting scores predict real-world outcomes. Each scenario represents a complete household profile; scores within ±3 of the target confirm correct behavior.
| Target | Actual | Tier | Household Profile | Storage Configuration | Rule Active |
|---|---|---|---|---|---|
| 0 | 0 | Low | Single adult, no stressors | Off-site storage | M=0, C=0 → Score=0 |
| 5 | 5 | Low | Veteran; suicide bereavement, financial, isolation | Off-site (gun shop) | M=0, C≥5 → Risk floor=5 |
| 10 | 10 | Low | Couple; financial stress, isolation, bereavement, job loss | Trigger lock, ammo not locked, status unknown | M=0.53, C=24 → Formula=10 |
| 15 | 15 | Low | Couple with children; domestic conflict, financial, isolation | Trigger lock, ammo not locked | M=0.53, C=35 → Formula=15 |
| 20 | 20 | Low | Couple with children; domestic, financial, isolation, mental health, job loss, bereavement | Trigger lock, ammo not locked | M=0.53, C=46 → Formula=20 |
| 25 | 25 | Moderate | Couple with children; suicidal ideation, domestic, financial, isolation, mental health, bereavement | Trigger lock, ammo not locked | M=0.53, C=58 → Formula=25 |
| 30 | 32 | Moderate | Couple with elderly parent; dementia, domestic, financial, substance, isolation, mental health | Hidden, ammo locked in separate location | M=0.67, C=46 → Formula≈30 |
| 35 | 35 | Moderate | Household with suicidal ideation and domestic conflict — ammo locked as protective factor | Hidden, unloaded, ammo locked (Rule 3 cap) | Rule 3: noLock+ammoLocked+isUnloaded → Cap=35 |
| 40 | 41 | Moderate | Couple; suicidal ideation, domestic, financial, substance, isolation, mental health | Hidden, ammo locked in separate location | M=0.67, C=59 → Formula≈40 |
| 45 | 45 | Moderate | Multi-generational; all major risk domains — ammo locked as only protective factor | Hidden, ammo locked (algorithm ceiling for this config) | M=0.67, C=65 → Formula=45 (max for hidden+ammo_locked) |
| ⚠️ Scores 46-64 are structurally unreachable — see explanation above | |||||
| 65 | 65 | High | Single adult, minimal stressors | Hidden, no lock — ammo accessible (Rule 1 minimum) | Rule 1: noLock+ammoNotLocked+hasRisk → Floor=65 |
| 70 | 70 | High | Couple; domestic, suicide bereavement, teen at risk, financial, isolation, mental health | Hidden, no lock — ammo accessible, last stored unloaded | M=0.91, C=55 → Formula=70 (Rule 1 floor exceeded) |
| 75 | 75 | Critical | Single adult, minimal stressors | Hidden, loaded — no lock (Rule 2 minimum) | Rule 2: isLoaded+noLock → Floor=75 |
| 80 | 81 | Critical | Couple with children; domestic, suicide bereavement, substance, financial, mental health | Hidden, loaded — no lock | M=0.93, C=61 → Formula=81 (Rule 2 floor exceeded) |
| 85 | 85 | Critical | Veteran couple; suicidal ideation, domestic, substance, mental health, financial, isolation | Accessible, loaded, within arm's reach (M=1.0) | M=1.0, C=55 → Formula=85 |
| 90 | 91 | Critical | Veteran couple; suicidal, domestic, dementia, suicide bereavement, financial | Accessible, loaded, within arm's reach (M=1.0) | M=1.0, C=59 → Formula=91 |
| 95 | 97 | Critical | Multi-generational; suicidal, domestic, dementia, substance, suicide bereavement, mental health, teen | Accessible, loaded, within arm's reach (M=1.0) | M=1.0, C=63 → Formula=97 |
| 100 | 100 | Critical | All risk factors and household types simultaneously present | Accessible, loaded, within arm's reach (M=1.0) | M=1.0, C=65 → Formula=100 |
The Structural Gap: Scores 46-64 Are Not Reachable
A key property of the FLAME v5 algorithm is that scores between 46 and 64 are structurally unreachable. This is intentional — it is a direct consequence of the floor/cap architecture.
The gap arises from the interaction of two rules:
- The highest score achievable with locked ammunition and unlocked storage is 45 — produced when a hidden, unlocked firearm has load status unconfirmed, ammunition is locked, and every other risk domain is at maximum (C=65, M=0.67, formula=44.9).
- The lowest score when ammunition is accessible and the firearm is unlocked is 65 — Rule 1 fires on any household risk factor, regardless of how low the formula would otherwise produce.
The 20-point discontinuity is the algorithm's expression of a clinical boundary: if a firearm is unsecured and ammunition is accessible, the household is in a categorically different risk zone than one where ammunition is locked. No configuration lands between 46 and 64. Scores either stay at or below 45 (ammunition controlled) or jump to at least 65 (ammunition accessible, firearm unlocked).
Staff training implication: A household scoring 45 — hidden gun, ammo locked, maximum risk factors — sits at the edge of the gap. A single change (switching from locked to accessible ammunition) causes the score to jump 20 points to 65. This transition is an explicit counseling target.
Computational Verification Table: 5-Point Increments
Each row is a complete, computationally verified household profile. All 18 reachable score points from 0 to 100 are covered. All scores verified against the live v5 algorithm to within ±2 points. This table confirms the algorithm behaves as designed; it is not a claim of predictive or clinical validity (see Validation Status). M = Accessibility Multiplier · C = Combined Risk Score (max 65).
| Score | Tier | M | C | Rule Active | Household Profile | Clinical Note |
|---|---|---|---|---|---|---|
| 0 | — | 0.00 | 0 | Formula (no risk) | All firearms off-site. No risk factors. Single adult, stress 1/10. | No means, no pathway. Score reflects absence of both access and risk. |
| 5 | — | 0.00 | 17 | Floor: off-site + risk = 5 | Veteran. Firearms at gun shop. Suicide bereavement, financial stress, isolation. | Risk floor of 5 applies — household risk factors remain clinically relevant even when firearms are off-site. |
| 10 | Low | 0.36 | 47 | Formula (no rules fire) | Couple with children/teen. Locked hard case, load unknown. Ammo accessible. Domestic, financial, job loss, housing, mental health, bereavement. | Locked case (M=0.36) compresses score despite moderate combined risk. Load status unknown prevents Rule 4 from firing. |
| 15 | Low | 0.44 | 48 | Rule 6 cap (≤35); formula=14 | Couple with children/teen. Locked case, confirmed unloaded, ammo accessible. Domestic, financial, mental health, isolation, bereavement. Stress 9/10. | Rule 6 cap is active but formula (14) lands well below it. Key recommendation: lock ammunition separately. |
| 20 | Low | 0.53 | 44 | Formula (no rules fire) | Couple with elderly parent. Trigger lock only, load unknown. Ammo accessible. Dementia, domestic, financial, mental health, job loss. Stress 6/10, impulse: locked. | Trigger lock (M=0.53) and reported locked impulse access moderate the score despite dementia and domestic conflict. |
| 25 | Low | 0.53 | 57 | Formula (no rules fire) | Couple with elderly parent and teen. Trigger lock, load unknown, ammo accessible. Dementia, domestic, substance, teen at risk, financial, isolation, mental health. Stress 7/10, hard: yes. | Same M as score 20 but higher combined risk (C=57 vs 44). Multiple urgent factors warrant counseling regardless of score. |
| 30 | Low | 0.61 | 53 | Formula (no rules fire) | Couple with children/teen. Trigger lock, confirmed unloaded, ammo accessible. Domestic, teen at risk, financial, isolation, mental health, housing. Stress 7/10, hard: yes. | Unlocked_unloaded anchor raises M to 0.61. Rule 6 does not apply — trigger lock is not a container lock. |
| 35 | Moderate | 0.75 | 54 | Rule 3 cap fires (≤35) | Couple with children. Hidden, no lock, confirmed unloaded. Ammo locked separately. Suicidal ideation, domestic, financial, isolation. Stress 7/10. | Rule 3 fires: hidden + ammo locked + unloaded = max 35. Formula would produce ~37; cap holds at 35. Locked ammo is the sole protective factor. Removing it jumps score to ≥65. |
| 40 | Moderate | 0.67 | 58 | Formula (no rules fire) | Couple with children and elderly parent. Hidden, no lock, ammo locked, load NOT confirmed. Domestic, dementia, substance, financial, mental health, isolation. Stress 7/10. | Load unknown prevents Rule 3. Confirming unloaded status would engage Rule 3 protection. Maintain locked ammo at all costs. |
| 45 | Moderate | 0.67 | 65 | Formula — ceiling for this config | Couple with children, teen, elderly parent. Hidden, ammo locked, load unknown. Suicidal ideation, domestic, dementia, autism/disability, substance, teen at risk, mental health. All domains maxed. Stress 10/10. | Algorithm ceiling for hidden + ammo locked + load unknown. FLAME=45 despite combined risk=65/65. Combined risk score must be reported alongside FLAME — staff must not treat 45 as "moderate" here. |
| STRUCTURAL GAP — Scores 46-64 are not reachable under the current algorithm. See section above. | ||||||
| 65 | High | 0.91 | 14 | Rule 1 floor fires (≥65) | Single adult. Hidden, no lock. Ammo in same location. Isolation, financial stress, mild mental health. Stress 3/10. | Rule 1 fires on minimal risk: no lock + ammo accessible + any risk = floor 65 regardless of formula (~5). The configuration is categorically dangerous. |
| 70 | High | 0.91 | 55 | Rule 1 satisfied; formula=70 | Couple with children/teen. Hidden, no lock, unloaded confirmed, ammo accessible. Domestic, teen at risk, financial, isolation, mental health, housing. Stress 7/10, hard: yes. | Same storage config as score 65. Formula (70) exceeds Rule 1 floor. The 5-point gap from 65 is driven by higher combined risk — not by any change in storage. |
| 75 | Critical | 0.93 | 7 | Rule 2 floor fires (≥75) | Single adult. Hidden, loaded, no lock. Ammo accessible. Financial stress only. Stress 2/10. | Rule 2 fires: loaded + no lock = floor 75 regardless of risk level. Ten-point jump from score 65 reflects additional lethality of loaded vs. unloaded unsecured firearm. |
| 80 | Critical | 0.93 | 61 | Rule 2 satisfied; formula=81 | Couple with children. Hidden, loaded, ammo accessible. Domestic, suicide bereavement (12 months), substance, financial, mental health, isolation. Stress 8/10, impulse: minute. | Suicide bereavement within 12 months (8 pts) is a significant driver. Rule 2 floor satisfied; formula drives score above floor. |
| 85 | Critical | 1.00 | 55 | Formula (M=1.0) | Veteran couple. Loaded, arm's reach, M=1.0. Suicidal ideation, domestic, substance, financial. Stress 8/10, impulse: seconds, hard: yes. | Maximum multiplier. FLAME = C/65 × 100 directly. Score=85 reflects C=55; adding child with confirmed access would push toward 90. |
| 90 | Critical | 1.00 | 59 | Formula (M=1.0) | Couple with elderly parent. Loaded, arm's reach. Suicidal ideation, domestic, dementia, financial. Stress 9/10, impulse: seconds, hard: yes. | Dementia (7 pts) is second only to domestic conflict (9 pts) among non-suicidal risk factors. Elderly household member raises combined risk to C=59. |
| 95 | Critical | 1.00 | 61 | Formula (M=1.0) | Veteran couple with elderly parent. Loaded, arm's reach. Suicidal ideation, domestic, dementia, substance, financial. Risk capped at 35. Stress 10/10, impulse: seconds. Child probe: yes, accessible. | Near-maximum. Risk domain at 35-point cap. 5-point gap from 100 reflects access domain not quite at ceiling. |
| 100 | Critical | 1.00 | 65 | Formula — all domains at ceiling | Multi-generational household. Loaded, arm's reach. All 14 risk factors active, all crisis indicators maxed. Child probe: yes, loaded. | Theoretical maximum. All three domains at ceiling (access=15, risk=35, crisis=15). Defines the scoring anchor. |
Key Design Innovations
Accessibility as Primary Score Driver
Placing accessibility as a multiplier — rather than an additive term — ensures the score reflects the causal mechanism: risk factors do not cause harm in isolation; they cause harm through access to lethal means. This architecture also produces more clinically meaningful score-impact estimates, because changing storage configuration changes the score dramatically and nonlinearly, mirroring the empirical evidence on means restriction effectiveness.
Behavioral Anchor Questions
Rather than asking respondents to characterize their general practices, FLAME asks about specific recent behaviors: "The last time you put a firearm away, what did you actually do?" Behavioral recall is more accurate and less subject to motivated distortion than self-characterization. Last-stored behavior adjusts the multiplier because it reflects what actually happened, not intent.
Vehicle and Outbuilding Storage
FLAME v5 explicitly captures firearm storage in vehicles and separate buildings. Firearms stored in vehicles present a distinct risk profile: they bypass household access barriers, are susceptible to theft, and may be accessible to individuals outside the home. The multiplier for unlocked vehicle storage (0.95) slightly exceeds that for home-accessible storage (1.0 maximum) to reflect this elevated unauthorized-access risk.
Suicide Bereavement as a Distinct Risk Factor
Previous versions of FLAME collapsed general bereavement and suicide bereavement into a single item. FLAME v5 separates these. General bereavement carries a weight of 3 points; suicide bereavement within the past 12 months carries a weight of 8 points — nearly as high as active suicidal ideation (12 points). This reflects the contagion literature, which identifies suicide bereavement — particularly within the first 12 months — as one of the strongest predictors of subsequent suicidal behavior in exposed individuals. (Jordan & McIntosh, 2011)
Projected Score Improvement
Each recommendation in FLAME v5 displays a projected FLAME score after that action is taken — not an abstract point reduction, but a specific new score. A summary box at the end of all recommendations shows the projected score if all recommended actions are followed. This approach leverages implementation intention research: concrete, quantified goals increase follow-through more than general behavioral intentions.
References
- Barber, C.W. & Miller, M.J. (2014). Reducing a suicidal person's access to lethal means. Am J Prev Med, 47(3S2), S264-S272.
- Crifasi, C.K. et al. (2018). Storage practices of US gun owners in 2016. Am J Public Health, 108(4), 532-537.
- Jordan, J.R. & McIntosh, J.L. (Eds.). (2011). Grief After Suicide: Understanding the Consequences and Caring for the Survivors. Routledge.
- Lubin, G. et al. (2010). Decrease in suicide rates after a change of policy reducing access to firearms in adolescents. Suicide Life Threat Behav, 40(5), 421-424.
- Miller, W.R. & Rollnick, S. (2013). Motivational Interviewing (3rd ed.). Guilford Press.
- Runyan, C.W. et al. (2016). Lethal means counseling for parents of youth seeking emergency care for suicidality. West J Emerg Med, 17(1), 8-14.
- Stanley, I.H. et al. (2017). Discussing firearm ownership and access as part of suicide risk assessment. Prof Psychol Res Pr, 48(4), 209-218.
- U.S. Department of Veterans Affairs (2024). VA National Veteran Suicide Prevention Annual Report. Office of Mental Health and Suicide Prevention.
- Wisconsin DHS (2023). Wisconsin Suicide Prevention Strategy 2023-2027. Madison, WI.