Safety documents

# Incident and near miss reporting

> Not legal advice. Which incidents must be notified to a regulator, how quickly, and what must be preserved at the scene varies by jurisdiction and is not optional. Find out what applies where you operate before you need to know.

## Near misses are the point

An injury tells you something went wrong. A near miss tells you the same thing, for free, before anyone is hurt.

A business that records near misses can fix the conditions that produce them. A business that only records injuries is waiting for one, and the wait is the only difference between the two.

**Near misses are only reported where reporting one is safe.** If reporting produces blame, a lecture, or paperwork the reporter has to complete, reporting stops within weeks. The count then falls, and the fall gets reported upward as a safety improvement, which is the exact opposite of what has happened.

**A falling near miss count is at least as likely to be a reporting problem as a safety improvement.** Check that interpretation first, every time, because it is the one nobody wants to be true.

## Reporting immediately

Memory degrades fast, and it degrades toward a tidier story than what happened.

- **Immediately for the facts.** A photo and three lines in the moment beat a careful report tomorrow
- **Make it two taps.** A form on the phone, against the job. Anything harder will not be used by someone who is dirty, tired and behind schedule
- **Anyone can report.** Not only supervisors
- **A verbal report counts.** Somebody else writes it up. Never make the reporter do paperwork as the price of reporting

```
servicem8_create_form_response   the report, from the device, against the job
servicem8_get_job_photos         photographs of conditions
servicem8_create_note            a quick verbal report written up by someone else
```

## What goes in the report

**Facts, in sequence. Not conclusions.**

```
WHAT HAPPENED
  At approximately 14:20 a 1.2m length of conduit fell from the level 2
  mezzanine to the ground floor slab. No one was beneath it. Two people
  were within 4m. The conduit was being passed up from the ground floor
  by hand.

CONDITIONS
  No exclusion zone below. Two other trades working at ground level.
  Conduit was being passed by hand because the hoist was in use on the
  south side.

WHAT WAS DONE IMMEDIATELY
  Work stopped. Area barricaded. Remaining conduit lifted by hoist after
  14:50.
```

Note what is absent: no judgement about who was careless, and no conclusion about cause. Both come later, from investigation, and putting them in the initial record contaminates it.

**"Why was it possible" gets somewhere. "Who did it" does not.** The second question ends the investigation at the first person and leaves every condition that produced the event exactly as it was.

## Investigating

Proportional. A near miss with no potential for harm is a note. One that could have killed somebody gets a real investigation whether or not anyone was touched.

**Rate by potential consequence, not by actual outcome.** A brick falling into an empty walkway and a brick falling onto someone's head are the same event with different luck. Investigate the first as seriously as the second, because next time the walkway will not be empty.

Ask why until you reach a condition rather than a person:

```
A conduit fell from the mezzanine.
  Why? It was being passed up by hand.
  Why? The hoist was in use elsewhere.
  Why? One hoist for two work fronts.
  Why? The programme has both fronts running in the same week.
  Why? The programme was compressed after the earlier delay.
```

The fix at the first level is "be more careful", which changes nothing. The fix at the fourth level is a second hoist or a resequence, which actually removes the hazard.

## Actions

A report with no action changed nothing. Each action needs an owner, a date, and a check that it happened.

| Action | Owner | By | Done |
|---|---|---|---|
| Exclusion zone below all mezzanine work, barricaded and signed | Ben | 16 Mar | |
| Second hoist hired for the duration of the dual front works | Dana | 18 Mar | |
| Toolbox talk on passing materials between levels | Ben | 15 Mar | |

**Close the loop with whoever reported it.** Tell them what changed. That single habit is the main determinant of whether the next near miss gets reported, and the next one might be the one that matters.

## What to keep and for how long

Varies by jurisdiction and is not a matter of preference. Find out the requirement locally. In general expect to keep incident records for years rather than months, and longer where injury or exposure to a hazardous substance is involved.

Practical habits regardless of the local rule:

- **Photograph the scene before anything is moved,** where it is safe and where nothing is required to be preserved
- **Record names of everyone present,** including other trades
- **Keep the original report**, not just the tidied summary
- **Record what the person said at the time,** in their words

## For a serious incident

Some incidents must be notified to a regulator, often within hours, and the scene may have to be preserved undisturbed. The thresholds and the timeframes are jurisdictional, they are strict, and finding out during the incident is too late.

**Establish now:** what must be notified, to whom, how fast, and who in the business makes that call. Put it on one page, put the phone number on it, and make sure the supervisors have it.

The parts that are true everywhere:

1. Make the area safe, and get medical help
2. Do not disturb the scene beyond what safety requires
3. Notify the person in the business who owns this decision
4. Notify the regulator if the threshold is met
5. Record everything as soon as practicable
6. Get advice before making statements about cause

## Reading the numbers

| Number | What it actually tells you |
|---|---|
| Near misses reported | Whether reporting works. Higher is usually better |
| Injuries | A lagging indicator. Too rare in a small business to show a trend |
| Actions closed on time | Whether reports lead anywhere |
| Repeat events of the same type | Whether the fixes are real fixes |

**Do not set a target of zero reported near misses.** It is the most reliable way to stop hearing about them, and businesses have done it, and it works exactly as badly as it sounds.

# Risk assessments and method statements

> Not legal advice. What must be produced, by whom, and kept for how long varies by jurisdiction and industry. Confirm the requirement locally.

## The difference between the two

| | Risk assessment | Method statement |
|---|---|---|
| Asks | What could go wrong and what do we do about it | How exactly do we do this task safely |
| Covers | A job, a site or an activity | One specific high risk task |
| Shape | Hazard, risk, control | A numbered sequence of steps |
| Read by | Supervisor, and the crew | The crew doing that task, before it |

Known variously as a SWMS, a safe work method statement, a method statement, a JSA or a job safety analysis, depending on where you are. The name matters legally and the content is much the same: the sequence of steps, the hazard at each step, and the control at each step.

**A method statement is not a risk assessment with steps.** Its value is entirely in the sequence, because sequence is where high risk work goes wrong: isolating after opening rather than before, removing the guard before the machine stops.

## The four steps

### 1. Identify hazards

Walk the job, in order, and ask what could cause harm at each point. Do it with the people who do the work, because they know the hazards that are not in any template.

Prompts that surface real ones:

- What has nearly gone wrong here before
- What is different about this site from the last one
- What happens if the power comes back on, the load shifts, the weather turns
- What is above, below and next to the work. Other trades especially
- What if this takes longer than planned and finishes in the dark
- Who else is on site who does not know we are here

**Do not work from a checklist alone.** A checklist catches the standard hazards and misses the one specific to this job, and the specific one is the one that hurts somebody.

### 2. Decide who could be harmed and how

Not just your crew. Other trades, the client's staff, the public, cleaners, anyone on site after you leave.

"How" matters as much as "who". The control for "struck by falling object" is different from the control for "trips over lead", and a row that just says "injury" has not been thought about.

### 3. Choose controls, from the top

```
1. Eliminate      remove the hazard.            Prefabricate at ground level
2. Substitute     something less hazardous.     Water based product
3. Isolate        separate people from it.      Lock out, barricade, exclusion zone
4. Engineering    design it out.                Guard, platform, extraction
5. Administrative change how people work.       Permit, procedure, training
6. PPE            protect the person.           Harness, gloves, glasses
```

**Levels 1 to 4 work without anyone remembering.** Levels 5 and 6 depend on a person doing the right thing, every time, under time pressure, at the end of a long day. That is why the order exists and why it is the most important idea here.

Most assessments go straight to 5 and 6 because they are cheap and nothing has to change. For every row, ask explicitly whether 1 to 4 were considered. Often level 1 was available and nobody asked.

**Controls must be specific enough to follow.**

| Not a control | A control |
|---|---|
| Be careful | Exclusion zone 3m, barricaded, signed |
| Use appropriate PPE | Class 1 insulated gloves, tested within 6 months |
| Ensure area is safe | Isolate at the main switch, lock out, test dead, tag |
| Work safely at heights | Scaffold with edge protection, erected by a licensed scaffolder, inspected before each shift |

### 4. Record, brief, review

- **Record it**, in a form the crew can open on site on a phone
- **Brief it.** An assessment nobody has read is not a control. Sign on means "I have read this and understood it", not "I was present"
- **Review it** when anything changes, and after any incident or near miss

## Rating risk

A matrix helps rank what to deal with first. It is a prioritisation tool, not a measurement, and precision beyond this is false.

| Likelihood \ Consequence | Minor | Moderate | Major |
|---|---|---|---|
| **Likely** | Medium | High | Extreme |
| **Possible** | Low | Medium | High |
| **Unlikely** | Low | Low | Medium |

Rate **before** controls and **after** controls. The after rating is what says whether the controls are adequate, and a row that is still Extreme after controls means the work does not start until something changes.

**Do not let scoring replace thinking.** A row scored Medium that everyone quietly agrees could kill someone has been scored wrong, and the score is now doing harm by providing false comfort.

## What a good one looks like

```
STEP        Remove and replace ceiling mounted light fittings, level 2 office

HAZARD      Fall from height, 3.1m to finished floor
WHO         Electrician, and anyone below
BEFORE      Unlikely / Major = Medium
CONTROLS    1. Scissor lift, not a ladder (engineering)
            2. Operator holds a current licence, verified on site
            3. Floor area below barricaded, 2m from the lift, signed
            4. Office area closed to staff during the works, agreed with
               facilities, confirmed by email 12 March
            5. Harness attached to the lift anchor at all times
AFTER       Unlikely / Minor = Low

HAZARD      Live circuit at the fitting
WHO         Electrician
BEFORE      Possible / Major = High
CONTROLS    1. Isolate at the distribution board, lock out with a personal
               lock, tag with name and date
            2. Test dead at the fitting with a tested instrument, before touching
            3. Test the instrument before and after, on a known source
            4. Keys held by the person working, not by the supervisor
AFTER       Unlikely / Major = Medium
```

Note what this does: the controls are mostly from the upper half of the hierarchy, each is specific enough to follow or to fail visibly, and one of them references a real email on a real date.

## Method statement sequence

For high risk work, write the sequence, because sequence is what goes wrong.

```
1. Confirm the permit is issued and current
2. Isolate at the main switch. Lock out with a personal lock, tag
3. Test the instrument on a known live source
4. Test dead at the point of work
5. Re test the instrument on a known live source
6. Apply earths where required
7. Begin work
```

Steps 3 and 5 are the ones that get dropped under time pressure, and they are the reason the sequence is written down. A step that is obvious to an experienced person is exactly the step a tired person skips.

## Reviewing someone else's

Five questions, in order:

1. **Is it about this job, or is it a template with the name changed?** Look for site specific detail. Its absence is the finding
2. **Are the controls from the top of the hierarchy, or is every row PPE?**
3. **Is each control specific enough to follow, and to fail visibly?**
4. **Is anything still high or extreme after controls?**
5. **Has the crew actually read it,** and is the sign on genuine?

Number 1 answers itself in about ten seconds, and it is the most common problem by a wide margin.

# Toolbox talks

Five to ten minutes, on site, one topic, with the crew who are about to do the work.

The format is old and it works, and it fails in exactly one way: a supervisor reading a printed sheet aloud while nobody listens, then passing it round for signatures. That version consumes ten minutes a week and controls nothing.

## What makes one work

**One topic.** A talk covering six things covers nothing. Pick the one thing that matters today.

**Relevant to today.** The best topic is almost always the most recent real event: last week's near miss, the new hazard that appeared when the other trade arrived, the change to the sequence. A generic topic from a pack is worth very little, and the crew can tell instantly which one they are getting.

**Two way.** The value is usually in what a worker says, not in what the supervisor reads. Ask a question and then stop talking. "Has anyone had this happen?" and then silence for ten seconds gets more than any prepared content.

**Short.** Five to ten minutes. A twenty minute talk is a meeting, and attention has gone by then anyway.

**Before the work, not at the end of the shift.** A talk given at knock off, to get the signatures, is a record of nothing.

## Choosing the topic

In order of value:

1. **A near miss from the last week or two.** Specific, recent, remembered. The highest value topic available
2. **A change today.** New site, new hazard, another trade arriving, weather
3. **Something the crew raised.** Closing a loop from a previous talk
4. **A seasonal or recurring hazard.** Heat, storms, shorter daylight, holiday fatigue
5. **An incident at another business,** where it is genuinely relevant
6. **A standard topic,** when there is nothing better. This is the fallback, not the default

**Rotate who runs it.** A talk given by a worker rather than the supervisor changes the dynamic and gets better participation, and the person preparing it learns the most.

## The shape

```
1. What happened, or what is different today     1 to 2 minutes
2. Why it matters here, on this job              1 minute
3. What we are doing about it                    2 minutes
4. Questions and anything the crew raises        2 to 4 minutes
5. Actions, with names                           1 minute
```

Section 4 is the point of the exercise. Sections 1 to 3 exist to earn it.

## Recording it

Date, topic, attendees, what came up, actions. That is all.

Where jobs are managed in a field system, record it as a form against the job. The signatures are captured on the device, the record has a timestamp that is genuine, and nobody is filling in a sheet for a talk that happened on Tuesday.

```
servicem8_create_form           the talk as a repeatable form
servicem8_create_form_response  a completed talk
servicem8_list_form_responses   the record
```

**Record what came up, not just that a talk occurred.** The list of things crews raised over six months is one of the most useful safety documents a business can have, and almost nobody keeps it.

## Following up

This is what separates a talk that works from theatre.

If somebody raises something, it gets an owner and a date, and it gets closed in a later talk. "Last week Jase pointed out the lead across the walkway. It is now run overhead, done Tuesday."

**A talk that surfaces a problem and changes nothing teaches the crew not to raise things.** That is worse than never having asked, because it removes the only mechanism the business has for finding out about hazards the supervisor cannot see.

## Twenty topics worth using

Hazard specific:

- Working at heights: the ladder decision, and when it is the wrong tool
- Electrical isolation: test dead, and testing the tester
- Manual handling: the lifts people actually do, not the poster
- Confined space entry: who is at the hole, and what they do
- Hot works: the fire watch, and how long it stays after
- Plant and vehicle movement: blind spots on this site
- Asbestos: what to do when you find something unexpected
- Silica dust: on tool extraction and water suppression
- Trenching and excavation: what is under this ground
- Chemical handling: the safety data sheet nobody has read

Site and behaviour:

- Housekeeping: the trip hazards on this job right now
- Other trades above and below you
- The public and the client's staff walking through
- Fatigue: the last hour of a long day
- Heat, hydration, and knowing when to stop
- Working alone: check in arrangements
- New starters: who is showing them what
- The shortcut everybody knows about
- Reporting a near miss, and why nobody gets in trouble for it
- What to do when you are asked to do something unsafe

That last one is worth running properly at least once a year, because the answer has to come from the top and be believed.

# Incident and near miss report

<!-- Not legal advice. Which incidents must be notified to a regulator, how fast,
     and what must be preserved at the scene varies by jurisdiction and is not
     optional. Establish that BEFORE you need it.

     Fill in sections 1 to 4 immediately, while memory is accurate. The rest can
     follow. A photo and three lines now beat a careful report tomorrow. -->

## 1. What kind

- [ ] Near miss, nobody harmed
- [ ] First aid injury
- [ ] Medical treatment injury
- [ ] Serious injury or illness
- [ ] Dangerous incident, no injury
- [ ] Property or plant damage only
- [ ] Environmental

**Potential consequence, had luck run differently:** <Minor / Moderate / Major / Fatal>

<Rate by POTENTIAL, not by what actually happened. A brick falling into an empty
walkway and a brick falling onto someone's head are the same event with different
luck. Investigate the first as seriously as the second.>

## 2. When and where

| | |
|---|---|
| Date and time | |
| Site and exact location | |
| Job number | |
| Reported by | |
| Reported to | |
| Time reported | |

## 3. What happened

<FACTS, in sequence. Not conclusions, and not judgements about who was careless.
Both come later, from investigation, and putting them here contaminates the record.>

## 4. Conditions at the time

<Weather, light, noise, who else was working, what plant was running, what was
different from normal, what pressure the job was under.>

## 5. People

| Name | Role | Company | Involved or witness | Injured |
|---|---|---|---|---|
| | | | | |
| | | | | |

<Include other trades and anyone from the client. Record what people said at the
time, in their words.>

## 6. Injury, if any

| | |
|---|---|
| Who | |
| Nature of injury | |
| Body part | |
| Treatment given | |
| Treated by | |
| Taken to | |
| Returned to work / sent home / hospital | |

## 7. Immediate action taken

<What was done in the first minutes. Work stopped, area secured, plant isolated.>

## 8. Photographs

<Photograph the scene before anything is moved, where it is safe and where nothing
is required to be preserved.>

- [ ] Scene photographed before anything moved
- [ ] Plant or equipment photographed
- [ ] Wider context photographed

## 9. Notification

- [ ] Regulator threshold checked against local requirements
- [ ] Regulator notified: <who, when, reference>
- [ ] Insurer notified
- [ ] Client notified
- [ ] Scene preserved as required

| | |
|---|---|
| Who made the notification decision | |
| Time | |

## 10. Investigation

<Ask why until you reach a CONDITION rather than a person. "Why was it possible"
gets somewhere. "Who did it" ends at the first person and leaves every condition
exactly as it was.>

| | |
|---|---|
| Why did it happen | |
| Why was that possible | |
| Why was that possible | |
| Why was that possible | |
| Underlying condition | |

**Was this foreseen in the risk assessment?**

- [ ] Yes, and the control failed. What failed: ________________
- [ ] Yes, but the control was not in place. Why not: ________________
- [ ] No. The assessment needs updating

## 11. Actions

<A report with no action changed nothing.>

| Action | Owner | By | Verified done |
|---|---|---|---|
| | | | |
| | | | |
| | | | |

- [ ] Risk assessment updated
- [ ] Method statement updated
- [ ] Toolbox talk delivered on this
- [ ] **Reporter told what changed**

<That last one is the main determinant of whether the next near miss gets
reported, and the next one might be the one that matters.>

## 12. Sign off

| | Name | Signature | Date |
|---|---|---|---|
| Reported by | | | |
| Investigated by | | | |
| Closed by | | | |

# Risk assessment

<!-- Not legal advice. What must be produced, by whom, and kept for how long
     varies by jurisdiction. Confirm the requirement locally.

     Fill this in for THIS job on THIS site. A generic assessment with the
     business name changed controls nothing and is visible in ten seconds. -->

| | |
|---|---|
| Job | |
| Site | |
| Date | |
| Prepared by | |
| Crew | |
| Reviewed with crew on | |
| Review due | <or: when anything changes> |

## The work

<Two or three lines. What is being done, where, and anything unusual about this
site compared to the last one.>

## Site specific conditions

<The section that proves this is not a template. Other trades, access, the public,
weather, what is above and below, what is different today.>

- <>
- <>

## Hazards and controls

<Controls from the TOP of the hierarchy: eliminate, substitute, isolate,
engineering, administrative, PPE. If every row is PPE, nothing has been
controlled. For each row, ask whether levels 1 to 4 were genuinely considered.>

| # | Step or hazard | Who could be harmed, and how | Before | Controls | After |
|---|---|---|---|---|---|
| 1 | | | <L/C = rating> | 1. <br>2. <br>3. | |
| 2 | | | | 1. <br>2. | |
| 3 | | | | | |

<Rating: Likelihood (Likely / Possible / Unlikely) and Consequence (Minor /
Moderate / Major).

Anything still Extreme or High after controls means the work does not start until
something changes.>

## Controls that must be in place before starting

- [ ] <>
- [ ] <>
- [ ] <>

## Permits required

- [ ] <Hot works / confined space / working at heights / electrical / excavation / none>

## Emergency arrangements

| | |
|---|---|
| Nearest hospital | |
| First aider on site | |
| First aid kit location | |
| Muster point | |
| Emergency contact | |
| Site address for emergency services | <The one to read out on the phone> |

## Sign on

<Signing means "I have read this and understood it", not "I was present".>

| Name | Signature | Date | Time |
|---|---|---|---|
| | | | |
| | | | |
| | | | |

## Review

| Date | What changed | Reviewed by |
|---|---|---|
| | | |

# Toolbox talk

<!-- Five to ten minutes, on site, ONE topic, with the crew about to do the work.
     Before the shift, not at knock off. The best topic is almost always the most
     recent real near miss. -->

| | |
|---|---|
| Date | |
| Time | |
| Site | |
| Led by | <Rotate this. A talk led by a worker gets better participation> |
| Topic | <One thing> |

## What happened, or what is different today

<1 to 2 minutes. Specific and recent. Last week's near miss beats any generic
topic, and the crew can tell instantly which one they are getting.>

## Why it matters here

<1 minute. On this site, today, to these people.>

## What we are doing about it

<2 minutes. Specific controls, not "be careful".>

- <>
- <>

## What the crew raised

<The point of the exercise. Ask a question and then stop talking. Ten seconds of
silence gets more than any prepared content.>

| Raised by | What | Action | Owner | By |
|---|---|---|---|---|
| | | | | |
| | | | | |

## Closing the loop from last time

<What was raised previously and what changed. A talk that surfaces a problem and
changes nothing teaches the crew not to raise things.>

- <>

## Attended

| Name | Signature |
|---|---|
| | |
| | |
| | |
| | |

---

## Topic ideas, when there is nothing better

<This is the fallback, not the default.>

Hazard specific: working at heights and the ladder decision | electrical isolation and testing the tester | manual handling | confined space entry | hot works and the fire watch | plant movement and blind spots | asbestos, finding something unexpected | silica dust | trenching and what is underground | chemical handling

Site and behaviour: housekeeping and today's trip hazards | other trades above and below | the public and client staff | fatigue in the last hour | heat and hydration | working alone and check ins | new starters, who is showing them what | the shortcut everybody knows about | reporting a near miss and why nobody gets in trouble | what to do when asked to do something unsafe