I am up to my ears in online advice that states correct that PPE should be based on a risk assessment basis.  Said risk assessment often being conducted on a standard grid based system.  I have also read everything I can find on what constitutes acceptable risk.  I also don't think you can ever eliminate risk entirely.  Which is indicated by any grid based assessment as they all have box you still stick a number or colour in for minimum risk and various degrees of hazard.  All the systems say do a risk assessment, but not much is written about where you draw the goal posts as a result.  So here is a scenario based on that grid:

Should you over - engineer PPE requirements and base it on the possibility of force majeure incident occurring?  Or should you set PPE based on expected daily conditions.  For example, if a factory manufacturing chemicals has a room with a number of pressurised lines in it that workers have to vist, and environmental monitoring show no contamination on a regular day and the plant is well maintained etc, should those workers on a daily basis be required to wear basic overalls, safety boots, maybe a hard hat, and safety specs in an everyday situation, or should the workers be required to wear a full impervious chemical suit and filter mask in case one of the lines gives way, or basic human error causes a release of something while they are in there on the basis that, if the worst happens, they are better protected?

I actually doubt the possibilty of coming up with a one size fits all chemicals PPE - you only have to look at all the different filter and glove type to realise the impracticality of such an approach.  However, I am quite interested in what degree of risk you factor into the assessment - the everyday or the extreme?  This morning I drove to work in a standard car and seatbelt, wearing everyday trousers and jumper.  Should I have been using a 4 point harnes and asbestos based undergarments in case I crashed into a petrol tanker.  Does the thinking change legally because you are in a work place?  Should you over-engineer PPE requirements just in case the other control measures in the risk assessment pyramid fail.  PPE isn't often the most comfortable thing to wear, it also might cause people to think 'I'm protected other things don't matter',  Is it better to plan for the worst and hope it doesn't happen?  Where do you stand legally?  etc.I will hold up my hand - I don't have much IOSH training, but at the moment I think I know where I stand.   I am really interested in the thoughts of those HSE IOSH pros with years of experience behind them.  Is it the hazard or the risk that is really important and when we answer that they are both important (which I know they are), where do you draw the line and make the decisions?  Also, does the advice on a SDS have any legal standing if that is all you decide to deploy?  I suspect it all tied up with Exposure scenarios, but we all know not everything that should, actually has those atm.

JOC1 - as you probably realise this all comes down to assessing what is and is NOT reasonably practicable.

Part of that assessment should include assessing the risks that the PPE will INCREASE.

So, you dress up Worker as Michelin Man or Michelle Woman and they end up suffering much worse effects from heat stress than the protection you are trying to give them.

Very rarely is this level of PPE reasonably practicable but lots of organisations do it anyway often as they have "Golden Rules" or whatever they like to call them.

....sometimes inherited from a US-like approach of ignoring the "general principles of prevention" aka "hierarchy of control measures" and starting from the bottom upwards.

What we should (in my opinion) be trying to do is to risk assess OUT of the need for any PPE, or at the very least to keep its need to the minimum. (Apologies to PPE suppliers!)

Next you need to decide (as Brian has alluded to) whether you are looking at the worst case consequence (almost invariably at least one dead body - even for someone just walking along the street, whether or not "at work") or some concept of a reasonably forseeable event.

Those who go for the worst case almost invariably overstate the reasonably predictable outcome EXCEPT when it comes to occupational health risks.

So, yes someone driving at work could be killed or cause someone else's death but it actually happens very rarely (in statistical terms) on UK roads and, (with limited exceptions) is more likely PER "passenger" mile to be the outcome on roads with speed limits of 40mph than on "high speed" roads, i.e. those with a speed limit of 50mph or more. 

Assuming that you have done some relatively basic health and safety training you should know that deciding on what is "reasonably practicable" involves balancing the risk against the cost (time, money and effort) of any mitigation being considered.

Which in turn brings us to the elephant in many rooms. Can you justify putting a value on someone's life or injury/ill health, to which the official answer is YES.

The UK Department of Transport has its VPF (Value for Preventing a Fatality) and various VPIs which are the values for preventing an injury, illness ete.

VPF is currently set at just over £2m - that is how much the Government assesses the socio-economic cost of a death e.g. on the roads.

A worker breaking a leg at work has a VPI of about £200,000.

The Rail Safety and Standards Board goes further in its Safety Risk Model and has numbers to tell you how many incidents of a certain severity make up one equivalent fatality.

So, if for example 1000 (this is a number taken out of the ether NOT what the Safety Risk Model says!) people are recorded as having fallen off the edges of station platforms each year very few will have escaped injury (mostly as they will not have been recorded), many will receive very minor injuries, some will sustain serious injuries and one or two might die.

Which means that you can work out the total socio-economic cost per annum and decide whether or not to retrofit interlocked barriers to stop people falling off each and every station platform when a train is not in place - and that decision might vary depending on the platform and its volume of traffic (amongst other variables).

All YOU have to do is the same but at a micro level, but with some guidance that may help you determine the package of precautions that might be proportionate to different work scenarios.

I didn't say it was easy.

Once you have done it, then possibly time to move on to considering the difference between the judgments on "reasonable practicability" in Edwards v NCB and Marhall v Gotham and to see what HSE has to say in R2P2.