We conduct a dig deeper for incidents that happen more than once to get to the root cause, but you do need to involve the affected parties and have them be open minded during the process. This may go a way to preventing any future occurrence and gets them engaged in the deeper thought process around incidents.

We run on a system where we look at the possible outcome risk of any incident and then investigate accordingly low level for "hitting finger with hammer" right up to a full ICAM for potentially major risks

Change the toolbox to behaviour based discussion(works best in team environment)

What be behavior led to the to the incident?
What behavior needs to change to ensure the incident does not re-occur.

The team member who had the accident and the team they in need to do before and after photo's and present the topic at the monthly toolbox bbq's.

From my perspective I think there are a several things to consider here.
Investigation process - why not get a member of the H&S committee involved in the investigation, investigations are best reviewed by a team of at least 2, so the line manager and the H&S committee member might be a good combination.
Outcomes - when the investigation is conducted is it looking at what was put in place previously and considering the effectiveness of those preventative measures? if you're having repeat occurrences, it may be that the measures have simply not been robust enough or even may not have tackled the causal factors.
Communication - it sounds as if your formula has been to integrate changes and results in toolbox talks, but repeating messages like this does have a negative effect, perhaps try an alternate mechanism - brief the team on the incident and ask them what would work better, get the person involved to talk about what they would do differently.
Trend analysis - sounds as if you have a trend, what is your company's approach to trend analysis and considering the wider implications? Consider all aspects of the work environment, and again, engage with team on what they need in order to stop the incident happening again.
Finally, in terms of engagement, toolbox talks absolutely have their place, but perhaps alternative mechanisms might be considered in addition, different people absorb information in different ways and at different rates, so try to have several ways of sharing the message.
Hope that's helpful

we are considering having different depths of investigations and feedback dependent on the incident, i.e. basic and full. — Chrissy Roff

We do something very like this - basic investigation is built in to every incident report, including injuries and near misses; then the responsible supervisor/manager/health & safety rep will decide if a "full" investigation is required. This is based on the actual or potential harm (hitting thumb with hammer would not rate very highly here), and the equipment/tools/scenario involved (we already know what our major risks are, so we would focus on these for a "full" investigation). Full investigations require consultation with staff, feedback through the health & safety committee, in-depth information gathering etc.

Hi Crissy
I would definitely recommend a tiered approach to investigations that not only relate to your company's risk appetite, but also your strategic goals. Let me know if you want an example procedure that outlines what I'm talking about and I'll hunt something down for you.
Regards
Cathy

Kia ora Chrissy

Our investigation system, relies on the input from the person whom is entering the incident on to our database. 9 times out of 10, this person is a H & S representative. When we get to the risk assessment section, the investigation will either be a 'simple' or 'full' investigation, dependant on the input. The system automatically allocates the type of investigation, dependant on the severity marking attributed to the incident.
Nga mihi
Grahame

We triage our incidents through the incident system we have (there are a set of system rules in behind that determine the level of investigation through the safety outcome, potential outcome and the risk classification of the incident.) This means that many (over 50%) require no investigation, with most of the remainder being learning team processes and a smaller amount of formal ICAM investigations.

The introduction of the learning teams process has been highly successful with us because it involves the teams that had the incidents to be involved. The teams enjoy this and the opportunity for learning is far greater.

This means that many (over 50%) require no investigation — Mike Massaar

What happens to these that don't get looked at (and what is an example of the worse case that wouldn't be investigated)?
And are these seen as just a normal part of doing business that can't really be avoided? And is there any review of trends in the non-investigated reports?

Very little. The organisation I work for covers 30% of NZ land area and we get lots of minor injuries such as twisted ankles, abrasions, small cuts, knocks and bruises etc etc. They are 80% of our incidents - and we report quite well. We just can't spend all that time investigating these. A broken bone would almost certainly always get investigated, but possibly at the learning team level, not a full ICAM investigation unless the incident had potential for something greater. .

While our view is that getting hurt working for us is not acceptable, reality is that many can't be avoided and for most there is little to learn from. We have a very good reporting system based around risk classifications, part of body, age, gender, work task, actual safety outcome and risk potential etc etc so we can generate very good and useful reports. For those low consequence incidents we do report back to those districts with recommendations on injury reduction. Despite our increasing staff numbers we have reduced these injuries reasonably significantly, but our focus is on the big stuff, and this has really hurt us in the past.

I investigate very few incidents. Most incidents are "minor" - that is a first aid type treatment or less. And most are due to the same cause. So there is little to be learnt from a new investigation, we have some inherent risk that we can't change and our management of those risks will always be proportional to the potential (and realistic) outcome. So basically I just wear first aid type injuries.

The ones that get my attention are the "near hit" ones. They pretty much consistently show something has continuously got under the radar and not been managed. So they need a fair bit of attention to remediate.

Doctor visit type incidents get a bit of attention - but each of these is on a case by case basis. We'll sometimes send a person to the doctor purely for a precautionary lookover and confirmation nothing major has happened.

Hospitalisations would get an investigation. But these are a bit trickier. Pretty much the only time ambulances are called out is due to personal health issues. So the "investigation" is more around is it right for the person to be remaining at work, given their health issues.

In a previous place of work I used the risk matrix with only the highest rated consequence factor to correspond to investigation level, i.e. Top 2 severities warranted a full formal investigation, the bottom two effectively an incident report review with the person/s involved to identify lessons learned and opportunities for improvements and the middle one (with a cross over to the second top) had a PEEPO and 5 whys type investigation. Investigation output should reflect actual or potential consequence. For this business, I didn't just focus on injury consequence but also environmental, reputational, property damage and regulatory (as the regulator had the ability to impact on operational capability).
Worked well, especially for instances if people say "why was I investigated but so and so wasn't...removes a potential for bias.

an example would be great thanks Cathy

I look after safety reports for a primary school, and we actually look for certain types of reports.
trips/grazes, bumps, minor sprains, etc are all a sign that the children are outside, active, and learning physical awareness.
A bit counter-intuitive, but not all injuries are bad, especially in a learning environment. Sometimes when people are physically pushing themselves, injuries are unavoidable.

On a previous role the trend analysis showed up a similar low level trend which led to talking with the people at the coalface about why it was occurring. We ended up swapping out tools which stopped the events from occurring again.

After the solution came from the coal face it was sent up the ladder and then back down to the toolbox talk to confirm the solution has been instigated which allowed the people at the coalface to see the positive effect they were having on the system.

Perhaps the order of who is discussing solutions should start with the people who it is firstly affecting.

Hi Chrissy,

I've created systems like this in the past, where the scale of the investigation is dependent on the specific incident.

For me, the key is that the depth of the investigation is determined by:
a) what the outcome could reasonably have been, not the actual outcome; and
b) how much can be learnt from the event (not always easy to assess without getting into it though)

It can be uncomfortable to spend a lot of time investigating an event when nobody got hurt, and even more uncomfortable to spend little time investigating something that did result in an injury - but sometimes that is the most logical approach! Otherwise, we are letting luck dictate how carefully we look at events. We end up spending tonnes of resource investigating things we have little control over and which are highly unlikely to cause harm in the same way again.... OR.... we end up overlooking high-potential events which we could learn heaps from, simply because luck smiled on us and thankfully nobody was injured. Lose lose.

When it comes to deciding how much there is to be learnt from near misses, it can help to ask, "Did nobody get hurt [or was the harm only minor] because our controls worked effectively.... or because we were lucky?"

I hope that is useful.