--- title: "Injury rates in Canada: lost-time claims, frequency and severity" description: "How Canadian injury rates are calculated — lost-time claim rate, LTIFR, severity rate — what counts, how experience rating prices it, and what does not compare." lede: "Canada has no single national injury rate, because the numerator is defined by whichever workers' compensation board covers you. That makes the headline number an insurer's measure rather than a safety measure — and knowing the difference is what stops a board meeting going wrong." jurisdiction: "Canada" locale: "en-CA" publishedOn: 2026-09-02 reviewedOn: 2026-09-05 canonical: "https://safeally.net/en-ca/guides/audits-and-records/lost-time-injury-rates/" sources: - label: "Data and Statistics — National Work Injury, Disease and Fatality Statistics Program" url: "https://awcbc.org/data-and-statistics" publisher: "AWCBC" - label: "Health and Safety Excellence program — learn about the program" url: "https://www.wsib.ca/en/healthandsafety" publisher: "WSIB Ontario" - label: "Canada Occupational Health and Safety Regulations, Part XV — Hazardous Occurrence Investigation, Recording and Reporting" url: "https://laws-lois.justice.gc.ca/eng/regulations/SOR-86-304/index.html" publisher: "Justice Laws Website" - label: "Incident Investigation — OSH Answers" url: "https://www.ccohs.ca/oshanswers/hsprograms/investig.html" publisher: "CCOHS" --- # Injury rates in Canada: lost-time claims, frequency and severity Canada has no single national injury rate, because the numerator is defined by whichever workers' compensation board covers you. That makes the headline number an insurer's measure rather than a safety measure — and knowing the difference is what stops a board meeting going wrong. ## The numerator is defined by your insurer, not by you This is the fact that makes Canadian injury rates different from European ones, and everything else follows from it. The headline number in Canada is the **lost-time claim rate**, and a lost-time claim is an **allowed claim** — one the workers' compensation board accepted, in which the worker lost time beyond the day of the accident, lost wages, or received a permanent disability award. Three consequences: - **A real injury whose claim was denied does not appear.** The board's adjudication decision, not your investigation, decides whether the event is in your rate. - **The definition varies by board.** Ontario's WSIB, WorkSafeBC, Alberta's WCB and the CNESST each publish rates using their own criteria and their own base. - **The number lags.** A claim allowed after a lengthy adjudication lands in the year of the accident, so a rate published early is provisional and will move. So a Canadian lost-time rate is an insurance measure that is used as a safety measure. It is still the number your board, your clients and your prequalification audits will look at — which means you need it, and you need at least one measure alongside it that you actually control. ## The formulas, and which base you are using **Lost-time claim rate per 100 workers** — the form most Canadian boards publish: > Rate = (number of allowed lost-time claims × 100) ÷ number of full-time > equivalent workers **Lost-time injury frequency rate (LTIFR) per 200,000 hours** — the internal management form: > LTIFR = (number of lost-time injuries × 200,000) ÷ total hours worked **Severity rate** — the days, not the count: > Severity rate = (days lost × 200,000) ÷ total hours worked The 200,000 is 100 full-time workers at 2,000 hours a year, so a rate per 200,000 hours and a rate per 100 workers are approximately the same magnitude. That approximation is exactly why the two get mixed up. **Total recordable rate** is the fourth one you will meet, and it is imported. Contractor prequalification systems used in Canada — ISNetworld, Avetta, ComplyWorks and the client audits behind them — often ask for a **TRIF** or **TRIR** built on US federal recordkeeping definitions, including medical-treatment cases with no lost time at all. If you fill that field with your Canadian lost-time number, you are understating against the definition asked for; if you fill it with a US-definition number, it will not match the rate you report to your board. Keep both, labelled, and say which is which. ## What counts, and what does not In the numerator of a lost-time rate: - allowed claims with time lost beyond the day of injury, or wage loss, or a permanent impairment award. Regularly and wrongly included: - **medical-aid-only claims** — real, reportable, and belonging in a separate count because their trend is a leading indicator; - **first aid only events** — not claims at all, but the most useful count you have for early warning; - **near misses** — not injuries, and a rise in reporting is good news rather than deterioration; - **occupational disease claims** — tracked separately, because latency makes attributing them to a year meaningless; - **commuting incidents** — generally not compensable as work injuries outside travel in the course of employment, and the rule differs by jurisdiction. In the denominator: **hours actually worked**, from payroll, including overtime. Three classic errors: | Error | Effect on the number | | -------------------------------------------------------- | ----------------------------------------------------- | | Headcount × 2,080 instead of actual hours | Misstates hours in whichever direction overtime runs | | Excluding contractor hours while counting their injuries | Rate artificially worsened | | Including paid absence and vacation hours | Inflates hours, so **artificially improves** the rate | The rule: the scope of the numerator and the denominator must be **the same**. If you count injuries to workers you direct, count their hours. If you cannot get their hours, do not count their injuries — and say so, because the number is then not the site's. ## Worked example Manufacturing site, 180 workers, 2025 calendar year. | Data | Value | | ------------------------------------------------- | ------------- | | Hours worked — direct employees | 262,400 h | | Hours worked — agency workers under our direction | 31,600 h | | **Total hours worked** | **294,000 h** | | Full-time equivalent workers | 200 | | Allowed lost-time claims | 6 | | Medical-aid-only claims | 9 | | First aid only events | 27 | | Near misses reported | 61 | | Calendar days lost | 214 | > **LTIFR** = (6 × 200,000) ÷ 294,000 = **4.08** per 200,000 hours > > **Lost-time claim rate** = (6 × 100) ÷ 200 = **3.00** per 100 workers > > **Severity rate** = (214 × 200,000) ÷ 294,000 = **145.6** days per 200,000 hours > > **Total recordable rate, US-style definition** = ((6 + 9) × 200,000) ÷ 294,000 = > **10.20** — the same site, more than double, purely because the definition > includes medical treatment > > **Average days lost per claim** = 214 ÷ 6 = **35.7 days** Two readings a board should be given together. First: the LTIFR of 4.08 and the recordable rate of 10.20 describe the same site — publishing one without saying which definition it uses is how a comparison goes wrong. Second: 35.7 days per claim is high, so this is a site with few but severe injuries. A site at LTIFR 8 with 15 days per claim has twice the injuries and half the severity, and its prevention priorities are different. The 61 near misses against 6 lost-time claims is a 10:1 ratio. Below about 5:1, under-reporting of weak signals is the likelier explanation than excellent safety. ## The number that actually costs you money Your lost-time rate is not what you are billed on. **Experience rating** is, and its mechanics differ by province in ways that change what prevention is worth. | Board | How injury history is priced | | --------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------- | | WSIB (Ontario) | Rate Framework: a prospective class rate, then a risk band adjusted by your own claims and insurable earnings history over several years | | WorkSafeBC | Experience rating on a multi-year claims-cost window, as a discount or surcharge against the base rate | | WCB Alberta | Industry rate plus an experience-based adjustment; **Partnerships in Injury Reduction** pays refunds of up to 20% for maintaining a Certificate of Recognition | | CNESST (Quebec) | Taux personnalisé for larger employers; smaller employers commonly join a **mutuelle de prévention** to share and reduce the rate | Three things worth telling a CFO, because they change the arithmetic: 1. **The lag runs both ways.** Experience rating windows are multi-year, so one bad claim is paid for across several premium years, and an improvement takes several years to show. Promising a premium effect next year is a promise you cannot keep, and it is what destroys the credibility of the argument the following year. 2. **Claims cost drives the rate, not claim count.** One long-duration or permanent-impairment claim can outweigh several short ones. This is the arithmetic behind modified work and early return to work — and also the reason it must be genuine: suitable modified duties reduce cost; pressuring an injured worker back reduces the claim and not the injury, and the boards look for exactly that pattern. 3. **Some rebates are paid for programme maturity, not outcomes.** Ontario's Health and Safety Excellence program pays per completed topic — a minimum of $1,000 or 100% of the prior year's premium per topic, whichever is less, up to $50,000. Alberta's PIR pays up to 20% for holding a COR. These are the only levers that pay in the current year rather than in three, which makes them the right first ask in a budget conversation. ## Reporting obligations, and their deadlines Getting these wrong costs faster than a bad formula. The specifics are provincial; the shape is national. | Obligation | Typical requirement | | ------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------- | | Worker reports the injury to the employer | As soon as practicable | | Employer reports the claim to the board | Commonly **3 days** of learning of a lost-time injury (Ontario Form 7); check your jurisdiction | | Immediate notice of a critical injury or fatality | **Immediately**, by telephone, to the regulator — and the scene is preserved | | Written report of a critical injury | Commonly within **48 hours** of the immediate notice | | Notification of the JHSC or representative | Immediately for critical injuries and fatalities | | Investigation and corrective action | Required; federally, COHSR Part XV prescribes the hazardous occurrence investigation, recording and reporting process | Two points that catch people: - **"Critical injury" is a defined term** with a specific list — it is not a judgement call, and misjudging it turns a reporting duty into an offence. Know your jurisdiction's definition before you need it. - **Preserving the scene is a legal duty, not a courtesy.** In most jurisdictions nothing may be disturbed at a critical-injury or fatality scene except to save life, relieve suffering, maintain essential services or prevent further injury, until the inspector releases it. Federally regulated employers additionally owe the **annual hazardous occurrence report** to the Labour Program, which is easy to overlook because it is annual and nothing prompts it. ## The indicators most dashboards are missing Lost-time rates are **lagging**: they describe what has already failed. A dashboard with two lagging numbers supports no decision before an injury. What belongs beside them: | Leading indicator | What it tells you | | --------------------------------------------------- | --------------------------------------------- | | Near misses reported per month, per area | Confidence in the reporting system | | Ratio of near misses to lost-time claims | Below 5:1, reporting has stalled | | Corrective actions closed on time (%) | Whether reporting leads anywhere | | Median days to close a corrective action | The real measure of how serious the system is | | Planned vs completed inspections and safety walks | Supervisory presence at the work face | | Hazard assessment entries revised after an incident | Whether the feedback loop exists | | Overdue PPE inspections and training expiries | Compliance debt, in countable units | | First aid events per 200,000 hours | The earliest reliable injury signal you have | A dashboard of eight indicators, six of them leading, is usable. A dashboard of two lagging indicators is an autopsy report. ## Why your numbers do not compare ### With a US site in the same group A US **TRIR** uses 200,000 hours and counts _recordable_ cases under US federal recordkeeping rules, which include medical treatment beyond first aid with no time lost. A Canadian lost-time rate counts allowed compensation claims with time lost. Same site, very different numbers — as the worked example above shows, roughly double. ### With a French site France uses a **taux de fréquence** per **1,000,000 hours**, counting accidents with at least one day off beyond the day of the accident, and its numerator is built from the accident declaration rather than from an accepted claim. The base differs by a factor of five and the numerator definition differs again, so scaling one into the other produces a figure that looks comparable and is not. ### With a German site Germany publishes a **1000-Mann-Quote** against _Vollarbeiter_ — full-time equivalents derived from hours — and counts only _meldepflichtig_ accidents, those causing more than **three days** of incapacity. The Canadian threshold is time lost beyond the day of injury. The German numerator is structurally smaller on the same site. **The only clean approach** in a multinational group: publish each jurisdiction's own measure with its definition attached, and compare each site's **trend against itself** rather than levels between sites. A leadership team accepts that answer readily when it is given before the question, and badly when it arrives after a comparison that had already convinced them. ## Sector specifics ### Construction The scope problem dominates. Hours worked by your employees, your agency workers and your subcontractors on one project sit in different accounts with different boards, and a "project" rate only means something if you state what it aggregates. Prime contractor and constructor duties mean you may be accountable for incidents that never touch your own claim experience. Track additionally: incidents by project phase, incidents by trade — which reveals gaps a project-level rate erases — and **Working at Heights training currency**, which in Ontario expires at three years and is a countable compliance debt. ### Manufacturing and industrial The sector where rates are most usable, because hours are known by line and by shift. Track additionally: - **incidents by operating state** — normal production, changeover, maintenance, shutdown. The distribution is almost always unfavourable to maintenance, and the site-level rate hides it; - **incidents by tenure in role** — the concentration in the first six months is large and drives orientation directly; - **incidents by shift and hour of shift** — nights and end-of-shift stand out, and those are scheduling levers, not awareness levers. ### Oil, gas, mining and resource Rates are commercial documents here, not just regulatory ones: client prequalification thresholds and bid eligibility attach directly to them, and an audited TRIF is routinely required alongside the provincial rate. Alberta's COR and PIR sit in the same conversation. Track contractor rates separately and transparently, because a blended rate that flatters the operator is the finding an auditor is looking for. ### Transportation and warehousing Jurisdiction is the first question: interprovincial carriers are federally regulated and report under COHSR Part XV, while a provincial warehouse reports to its board. A single company can be both. Motor vehicle incidents are the leading cause of workplace fatality in Canada, and they belong in the rate as occupational injuries. Add a rate per kilometre driven — the only meaningful denominator for road risk — and separate dock from non-dock incidents. ### Healthcare and long-term care The numerator has particular contents: musculoskeletal injury from patient handling dominates severity; workplace violence is heavily under-reported and should be counted separately with its own target; needle-stick and biological exposure events are reportable and worth tracking whether or not time is lost. Modified work is easier to arrange in a large healthcare employer than almost anywhere else, which means the lost-time rate understates the injury burden more here than elsewhere. Publish the medical-aid count next to it. ### Office and professional Rates are low and therefore statistically unstable: across 200 office workers, one claim moves the rate from 0 to 0.5 per 100, and tracking that monthly produces noise. Track instead: short-duration absence by team, mental-health-related claims and absences, musculoskeletal complaints related to workstation and remote work setup, and road incidents for anyone who drives to clients — often the only genuinely severe risk the organisation has. ## What an inspector or an auditor checks - [ ] **Claims were reported to the board within the required time.** - [ ] **Critical injuries and fatalities were reported immediately**, and the scene was preserved. - [ ] **The written follow-up report was filed** within the prescribed period. - [ ] **The JHSC or representative was notified and participated** in the investigation, as the act requires. - [ ] **Investigations identify causes beyond the immediate one**, and produce corrective actions. - [ ] **The hazard assessment was revised** after each incident. - [ ] **Federally: the annual hazardous occurrence report was submitted.** - [ ] **Indicator definitions are written down** — scope, numerator, denominator, hours source. - [ ] **The definition has not changed without being recorded** in the published history. - [ ] **Corrective actions have an owner, a due date and a status.** - [ ] **Reported rates match what was reported to the board** and to prequalification systems. A discrepancy between the two is the finding. The last two are the ones that surprise people. An indicator whose definition is not written is not auditable, and it will have changed at least once in three years without anyone remembering — almost always when the role changed hands. And different numbers reported to a client and to a board is a serious finding, even when both were computed honestly under different definitions. ## What to write **The definition note — one page, written once.** > **Lost-time injury frequency rate — definition, [Site], version 2026-01** > **Numerator:** allowed lost-time claims for direct employees and for agency > workers under our operational direction. Excluded: medical-aid-only claims, > first aid events, occupational disease claims, commuting incidents, recurrences > (attributed to the original claim). > **Denominator:** hours actually worked from payroll, direct and agency, overtime > included, paid absence and vacation excluded. > **Base:** 200,000 hours. > **Hours source:** monthly payroll extract, code [Y]. > **Publication:** monthly, 12-month rolling. > **History:** unchanged since 2024-01. Any change is dated here and the history > is restated. **The dashboard paragraph that prevents the wrong question.** > LTIFR, 12-month rolling: **4.08** (2024: 5.44). Severity rate: **145.6** (2024: > 102). Fewer claims, more severe: both claims over 60 days lost this year were > manual handling in packaging, area COND-02, which is the priority action in the > current plan. This is not comparable to the group TRIR — different definition and > base, see the definition note. Total recordable rate on the US definition for the > same period is 10.20. **Phrases to stop writing.** | What gets written | Why it is a problem | Write this instead | | ------------------------------ | ------------------------------------------------------------ | --------------------------------------------------------------- | | "LTIFR down 25%" | On 6 claims, one fewer is −17%: that is noise | "6 claims against 8: not a significant change at this exposure" | | "Target: zero injuries" | Unmeasurable at small exposure, and it pushes reporting down | "Target: 100% of corrective actions closed within 60 days" | | "Better than industry average" | Compares an LTIFR to a per-100-worker board rate | Compare your board rate to your board's sector rate, or nothing | | "Near misses up" | Presented as deterioration when it is progress | "Weak-signal reporting up 40%: the reporting system is working" | | "Safety bonus tied to LTIFR" | Pays for non-reporting | Index it to leading indicators, never to the injury rate | | "TRIR: 4.08" | A Canadian lost-time number in a US-definition field | Report both, each labelled with its definition | The bonus point deserves saying plainly, because it is widespread and counterproductive: **a crew bonus indexed to the injury rate buys silence, not safety.** The observable effect is a fall in reporting with no change in serious injuries — which do not hide. If a bonus must exist, index it to things the work face can improve by working better: reports raised, actions closed, inspections completed.