ohs-data-advisor

name

ohs-data-advisor

description

Australian WHS statistics and occupational hazard reference advisor for Safetysure consulting. Use when the user asks about occupational hazard profiles, injury mechanisms, industry injury rates, WCIFR benchmarks, national fatality or serious claims statistics, or WHS data for any occupation or industry. Three operating modes: (A) standalone hazard/statistics query — "what hazards do concreters face?", "what is the WCIFR for construction?", "what mechanisms cause most injuries?"; (B) hazard gap check — co-load alongside safetysure-swms or safetysure-audit-criteria to cross-check document controls against BOHD occupation hazard profiles; (C) risk assessment likelihood support — provides national data reference to inform likelihood ratings in risk registers and risk matrices for a given occupation/hazard. Also triggers on: "hazard profile", "BOHD data", "SWA statistics", "injury benchmark", "occupation risk", "national injury rate", "frequency rate", "incidence rate", "what injuries", "top hazards", "industry claims data", "likelihood rating", "risk register", "risk matrix", "assign likelihood", "what likelihood", "rate this hazard", "how likely is", "risk assessment data".

OHS Data Advisor — Australian WHS Statistics and Hazard Reference

Purpose

This skill provides Safetysure consultants with access to verified Australian WHS statistics and occupation-specific hazard profiles. All data sourced from Safe Work Australia (SWA) official publications. The skill supports:

• Hazard identification — what hazards are workers in a given occupation nationally exposed to, at what frequency?
• Gap analysis — what hazards are documented in official data but not addressed in a SWMS, risk register, or safety plan?
• Industry benchmarking — how does a client's injury rate compare to their industry peers?
• Report statistics — verified national figures for use in consulting reports
• Likelihood rating support — provides national data reference points to inform initial likelihood ratings in risk registers and risk matrices

Core limitation: All data is national averages. This skill identifies hazards that are statistically prevalent for an occupation or industry — it cannot assess the actual risk level at a specific site. Site-specific conditions may increase or decrease exposure relative to national averages.

---

Data Sources and Reference Files

Load reference files selectively based on query type. Do not load all files at once.

Reference file	Content	Load when
bohd-occupation-profiles.md	318 occupations × hazard exposure profiles + workers' comp stats	Any occupation-specific query or SWMS/audit gap check
national-hazard-overview.md	National fatality + serious claims statistics, mechanism breakdown, hazard exposure categories	National statistics queries, report writing, general risk context
industry-claims-data.md	Industry longitudinal fatality and serious claims data, all 19 ANZSIC divisions	Industry benchmarking, audit criteria prioritisation
injury-mechanisms.md	TOOCS mechanism, nature, and bodily location frequencies (NDS 2023-24)	Mechanism-specific queries, SWMS mechanism gap check
wcifr-benchmarks.md	Workers' compensation injury frequency rates by industry and occupation	Benchmarking queries, frequency rate comparisons, Mode C likelihood support

---

Operating Modes

Mode A — Standalone Query

Triggers: User asks directly about hazards, statistics, or benchmarks.

Workflow:

1. Identify query type:

   • Occupation hazard profile → load bohd-occupation-profiles.md
   • Industry claims / benchmarking → load industry-claims-data.md + wcifr-benchmarks.md
   • Injury mechanisms / how injuries happen → load national-hazard-overview.md + injury-mechanisms.md
   • National statistics for a report → load national-hazard-overview.md
   • Full hazard picture for an occupation in an industry → load bohd-occupation-profiles.md + injury-mechanisms.md + wcifr-benchmarks.md
   • Likelihood rating support (risk register / risk matrix) → see Mode C below — use bohd-occupation-profiles.md + wcifr-benchmarks.md + industry-claims-data.md

2. Find the occupation or industry in the reference data:

   • Search bohd-occupation-profiles.md by occupation title or ANZSCO code
   • If occupation not found at 4-digit level, use the ANZSCO major group profile as a fallback
   • Search industry-claims-data.md by ANZSIC division letter or industry name

3. Present findings in the Standard Profile Output format (see Output Formats below)

4. Always cite: source dataset name, vintage (e.g. "BOHD, SWA Dec 2023"), and the beta/preliminary caveat where applicable

---

Mode B — Hazard Gap Check (co-load with SWMS review or audit criteria)

When invoked: The safetysure-swms skill (swms-review) or safetysure-audit-criteria skill (criteria-generator or checklist-completer) loads this skill after identifying the occupation and/or industry from the document.

Workflow:

1. Receive from the calling skill:

   • Occupation title and/or ANZSCO code (from SWMS job title / work description)
   • Industry (ANZSIC division, if relevant)
   • List of hazards / controls already identified in the document

2. Load bohd-occupation-profiles.md and locate the occupation entry

3. Build the gap matrix:

   • Column A: BOHD hazards with exposure score ≥ 50 (nationally prevalent)
   • Column B: Whether each hazard is addressed in the document (yes / partial / no)
   • Column C: Exposure score (severity proxy)
   • Column D: Priority flag — P1 if hazard is ≥ 75 AND not addressed; P2 if ≥ 50 AND not addressed

4. Also check injury-mechanisms.md — if a nationally high-frequency mechanism (e.g. body stressing, falls) is not covered by any control in the document, flag it

5. Present findings in the Gap Check Output format (see Output Formats below)

6. Do not override the calling skill's findings — present this as supplementary statistical context, not a regulatory compliance determination

For audit criteria (criteria-generator co-load):

After generating the criteria checklist, load industry-claims-data.md and wcifr-benchmarks.md for the relevant industry. Annotate high-priority criteria items with the industry's top injury mechanisms and frequency rate. This helps the auditor focus attention on statistically prevalent risks for that industry.

---

Mode C — Risk Assessment Likelihood Support

Triggers: User is completing a risk register or risk matrix and asks for data to inform the likelihood rating for a hazard/occupation combination. Also triggered when a risk assessment calls for an evidence-based starting point for initial likelihood ratings.

Core principle: This mode provides national data as a reference starting point only. Likelihood ratings in a risk matrix are site-specific — the consultant must apply professional judgement and adjust for local conditions, controls already in place, and the organisation's risk matrix scale. National data informs the initial rating; it does not determine it.

Workflow:

1. Identify the hazard and occupation (and industry where relevant) from the risk register or user query

2. Load the relevant reference files:

   • bohd-occupation-profiles.md — BOHD exposure score for the hazard/occupation pair
   • wcifr-benchmarks.md — industry or occupation WCIFR (lost-time claims per million hours worked)
   • industry-claims-data.md — serious claims incidence rate (per 1,000 workers/year) for the relevant industry
   • injury-mechanisms.md — if the hazard maps to a specific TOOCS mechanism, load for national frequency context

3. Map the data to a likelihood reference band using the Likelihood Reference Table below. This produces a suggested starting band — not a prescribed rating

4. Present findings in the Likelihood Support Output format (see Output Formats below)

5. Always explicitly note:

   • The rating is a national data reference point only
   • Site-specific controls, supervision, worker experience, and environmental conditions may justify a different rating
   • The consultant must determine the final likelihood rating using their own risk matrix scale and site-specific assessment

Likelihood Reference Table

This table maps national data signals to a likelihood reference band. Bands are intentionally broad — national data cannot substitute for site-specific assessment. Use this as an anchor, not a formula.

Signal	Low reference	Moderate reference	High reference
BOHD exposure score	< 40	40–69	≥ 70
WCIFR (lost-time claims/million hrs)	< 5	5–15	> 15
Incidence rate (serious claims/1,000 workers/yr)	< 5	5–20	> 20
National mechanism share (% of all serious claims)	< 5%	5–20%	> 20%

Reading the table: Apply all available signals and identify the most common band across them. Discordant signals (e.g. high exposure score but low WCIFR) indicate that site-specific judgement is especially important — note this in the output.

Mapping to common risk matrix scales:

Reference band	5-point likelihood	4-point likelihood	3-point likelihood
High	Almost Certain / Likely (4–5)	Likely / Possible (3–4)	High (3)
Moderate	Possible / Unlikely (2–3)	Possible / Unlikely (2–3)	Medium (2)
Low	Rare / Unlikely (1–2)	Rare / Unlikely (1–2)	Low (1)

Important: These mappings assume no controls are in place (inherent likelihood). Where the risk register records residual likelihood (post-controls), adjust downward based on the effectiveness of controls documented for that hazard.

---

Output Formats

Standard Profile Output (Mode A)

**Hazard Profile: [Occupation Title] (ANZSCO [code])**
*Source: BOHD, Safe Work Australia, December 2023 (beta dataset — indicative)*

**Physical/Environmental hazards (score ≥ 50):**
[List hazard — score]

**Ergonomic hazards (score ≥ 50):**
[List hazard — score]

**Psychosocial hazards (score ≥ 50):**
[List hazard — score]

**PPE indicators:**
[List — score]

**Workers' compensation (national average):**
Serious claims: [N]/yr | Incidence: [X] per 1,000 workers | Frequency: [Y] per million hours

**Industry context (if relevant):**
[WCIFR benchmark for relevant ANZSIC division]

*All figures are national averages. Site-specific conditions may differ.*

Gap Check Output (Mode B)

**OHS Data — Hazard Gap Check: [Occupation] (ANZSCO [code])**
*Cross-referenced against BOHD (SWA, Dec 2023). Exposure score ≥ 50 = moderate-to-high national average.*

| Hazard | Exposure score | Addressed in document? | Priority |
|--------|---------------|----------------------|----------|
| [Hazard 1] | [score] | ✅ Yes | — |
| [Hazard 2] | [score] | ❌ No | P1 |
| [Hazard 3] | [score] | ⚠ Partial | P2 |

**P1 gaps** (score ≥ 75, not addressed): [list — these warrant immediate attention]
**P2 gaps** (score 50–74, not addressed): [list — consider adding controls]

*Note: This gap check identifies hazards that are statistically prevalent nationally for
this occupation. It does not replace a site-specific risk assessment — conditions at the
specific workplace may increase or decrease these exposures.*

Likelihood Support Output (Mode C)

**Likelihood Data Reference: [Hazard] — [Occupation] (ANZSCO [code])**
*Source: BOHD (SWA Dec 2023), WCIFR 2023-24, NDS 2023-24 (preliminary). National averages only.*

**National data signals:**
- BOHD exposure score: [X]/100 → [Low/Moderate/High] reference
- WCIFR ([ANZSIC division]): [Y] per million hours → [Low/Moderate/High] reference
- Serious claims incidence ([industry]): [Z] per 1,000 workers/yr → [Low/Moderate/High] reference
- [Mechanism] share of national serious claims: [N%] → [Low/Moderate/High] reference

**National data reference band: [Low / Moderate / High]**

**Suggested starting point (inherent likelihood, pre-controls):**
- 5-point scale: [range, e.g. Possible–Likely (3–4)]
- 4-point scale: [range]
- 3-point scale: [band]

**Notes for site-specific adjustment:**
[Any discordant signals, gaps in data, or factors that may warrant adjustment — e.g.
"High exposure score but moderate WCIFR — inherent hazard is significant but industry
controls appear effective at national average level"]

⚠ *This is a national data reference point only. Apply professional judgement and
adjust for site-specific conditions, controls in place, worker experience, and your
organisation's risk matrix scale before assigning a final likelihood rating.*

---

Integration Notes for Other Skills

safetysure-swms (swms-review)

After completing the SWMS compliance review, optionally co-load ohs-data-advisor in Mode B to add a BOHD-based hazard completeness check. This surfaces hazards that are statistically prevalent for the occupation but may not be visible from the SWMS document alone. Present as a supplementary section headed "Hazard Completeness Check (BOHD Data)".

safetysure-audit-criteria (criteria-generator)

After generating the base criteria checklist, optionally co-load ohs-data-advisor in Mode B to annotate criteria items with the industry's top hazard mechanisms and WCIFR benchmark. This helps prioritise which criteria items deserve the most attention during a site audit.

safetysure-ohs-stats (aioh-compliance)

If occupational hygiene exposure monitoring has been completed, ohs-data-advisor can provide national context: how many workers in this occupation are nationally exposed to this hazard (from BOHD), and the occupation's overall incidence rate. Present as contextual framing around the compliance determination.

---

Mandatory Caveats

Always include these caveats when presenting data from this plugin:

1. BOHD is a beta dataset — "Exposure scores are indicative national averages from the Beta Occupational Hazards Dataset (SWA, Dec 2023). Use for hazard identification and gap analysis — not as a definitive risk quantification."

2. National averages only — "All statistics represent national averages. Site-specific hazard levels may differ significantly from these figures."

3. Data vintage — Always cite the data year for any statistic (e.g. "2023-24 preliminary", "BOHD December 2023", "Key WHS Statistics 2025").

4. Compensation data scope — NDS and WCIFR data covers compensated claims only (≥1 working week lost). Non-compensated injuries, near-misses, and illnesses with shorter time off are not captured.