The hierarchy of controls, and how to actually use it

The hierarchy of controls is the order you are expected to work through when you decide how to deal with a hazard. It runs from the most effective control to the least, and the point of it is simple: get rid of the danger if you can, and only fall back on asking people to protect themselves when you have genuinely run out of better options.

It is not a menu. You do not pick the one that suits your budget. You start at the top and you only move down a level when the level above is not reasonably practicable. That phrase matters, because it is the test an HSE inspector will apply and the test a court will apply, and “we gave them gloves” is a very weak answer when a guard would have done the job.

The five levels, in order

• Elimination. Remove the hazard completely. Nobody can be hurt by something that is not there.
• Substitution. Swap it for something less dangerous.
• Engineering controls. Design the risk out, or put a physical barrier between the hazard and the person.
• Administrative controls. Change how the work is done: procedures, training, signage, permits, rotation.
• Personal protective equipment. Protect the individual, as a last resort and never as the only measure.

The first three act on the hazard. The last two act on the person. That is the real dividing line, and it is why the order is what it is. Controls at the top keep working whether or not anybody remembers to do anything. Controls at the bottom only work if a person does the right thing, correctly, every single time.

Where this sits in UK law

You will not find the words “hierarchy of controls” in the Health and Safety at Work etc. Act 1974. What you will find is the duty to ensure health and safety so far as is reasonably practicable, and a hierarchy is how you demonstrate you have met it.

Several sets of regulations do spell it out. Schedule 1 of the Management of Health and Safety at Work Regulations 1999 sets out the general principles of prevention, which begin with avoiding risks altogether and end with giving appropriate instructions to employees. The Control of Substances Hazardous to Health Regulations 2002 require you to prevent exposure first and only control it where prevention is not reasonably practicable. The Work at Height Regulations 2005 tell you to avoid work at height, then prevent falls, then mitigate the distance and consequences of a fall. The Control of Noise at Work Regulations 2005 make clear that hearing protection is not a substitute for reducing noise at source.

Different wording, same idea every time. If you are studying for the NEBOSH National General Certificate or IOSH Managing Safely, you will meet this in the first few days, because almost everything else builds on it.

Elimination: get rid of it

The most effective control, and the one most often skipped because it usually means changing the job rather than changing the kit.

• Prefabricate off site and lift into place, so nobody assembles anything at height
• Specify a quieter machine rather than issuing ear defenders around a loud one
• Design out the need to enter a confined space by fitting external sampling points
• Have the delivery come to the point of use rather than carrying it across site
• Dispose of the drum of solvent nobody has used since 2019

Elimination is cheapest at the design stage and most expensive once the job is running, which is why the CDM Regulations put so much weight on designers. If you are being asked to eliminate a hazard on a live site, you have usually inherited somebody else’s decision.

Substitution: swap it for something safer

If you cannot remove the hazard, replace it with a lesser one. Water based paint instead of solvent based. A pellet or granule instead of a fine powder that hangs in the air. A battery tool instead of a petrol one indoors. A lighter component so the lift is within a sensible range.

The catch is that a substitute brings its own hazards, and they have to be assessed properly rather than assumed to be better. Steam cleaning avoids the solvent but introduces a burn and a heat stress risk. A water based product may need a biocide that carries its own COSHH implications. Substitution is only a win if you have actually checked.

Engineering controls: design the risk out

Physical measures that work without anybody having to remember them. These are the workhorse of most real control plans.

• Guarding. Fixed and interlocked guards on moving parts, as required by PUWER
• Isolation. Putting distance, an enclosure or an acoustic booth between the hazard and the people
• Local exhaust ventilation. Capturing dust, fume or vapour at source before it reaches the breathing zone
• Edge protection. Guardrails, toe boards and covers so a fall cannot happen, rather than a harness for after it does
• Mechanical handling. Hoists, vacuum lifters, pallet trucks and conveyors instead of people carrying things
• Interlocks and trips. Machinery that will not run with the gate open

Engineering controls only stay effective if they are maintained. LEV needs a thorough examination and test at least every fourteen months under COSHH. A guard that has been propped open because it slows the job down is not a control, it is a defence you no longer have.

Administrative controls: change how the work is done

Procedures, permits, training, supervision, signage, job rotation, restricting who is allowed to do the work and when. These are ranked below engineering controls for one reason: the hazard is still there, and the control depends on human behaviour.

• Safe systems of work and method statements that describe the job as it is actually done
• Permit to work for hot work, confined spaces, live electrical work and roof access
• Planned preventative maintenance, so equipment fails safe rather than fails
• Scheduling high exposure tasks when the fewest people are on site
• Restricting a task to people who are trained, assessed and authorised for it
• Job rotation to limit how long any one person is exposed

Training belongs here, and it is worth being honest about what that means. Training is an administrative control. It is necessary, we sell a lot of it, and it is still the fourth best answer. If the only thing standing between your people and a serious injury is that they went on a course, you have more work to do.

PPE: the last line, never the only line

Personal protective equipment protects one person, only while they are wearing it, only if it fits, and only if it is the right type for the hazard. Under the Personal Protective Equipment at Work Regulations 1992, as amended in 2022 to cover limb (b) workers, PPE is what you use for the risk that is left after everything above has been applied.

Where PPE earns its place is as a supplement, not a substitute: the right PPE alongside a guard and a procedure, not instead of them. Respiratory protective equipment in particular needs face fit testing for tight fitting masks, a maintenance regime and somewhere clean to store it, or it is doing considerably less than the wearer thinks.

Most real jobs need more than one level

Hazards rarely get dealt with by a single control. A realistic plan usually stacks several.

Take cutting paving slabs on site. Eliminate where you can by ordering them cut to size. Where you cannot, substitute the dry cut for a wet cut to suppress the silica dust. Add on tool extraction (engineering). Restrict the task to trained operatives with a current abrasive wheels certificate and rotate the work so no one person does it all day (administrative). Then fit tested RPE and eye protection on top (PPE). Five levels, one job, and the dust never gets near anybody’s lungs.

Work at height is the same pattern. Avoid it, then prevent the fall with a tower or a MEWP and edge protection, then and only then mitigate with a harness and an anchor point. Our working at height awareness course works through exactly that sequence.

Applying it to a risk assessment

The hierarchy is not a separate exercise. It is the “what are we going to do about it” column of your risk assessment, and the reason that column so often looks thin is that people jump straight to the bottom of the list.

• Identify the hazard and who could be harmed, and how
• Work down the hierarchy in order, recording what you rejected and why
• Write down the controls you are actually implementing, not the ones you aspire to
• Put a name and a date against each one
• Check afterwards that the control works and has not created a new hazard
• Review when the work, the equipment, the people or the law changes

Recording what you ruled out is the part most people skip, and it is the part that protects you. “Elimination not reasonably practicable because the structure is listed and the opening cannot be moved” is a defensible position. A blank space is not. Our risk assessment eLearning covers how to write that up properly.

Common mistakes

• Starting at PPE. It is quick, it is visible and it is cheap, which is exactly why it is so often the first answer instead of the last.
• Counting training as a fix. Training supports a control. It is not one on its own.
• Fitting a control and never checking it. LEV that has never been tested, a guard that has been bypassed, a permit system nobody uses after 4pm.
• Creating a new hazard. Full body coveralls in July cause heat stress. A heavier guard causes a manual handling problem. Check what you have introduced.
• Assessing the job on paper. Go and watch it being done. The gap between the method statement and the reality is where most incidents live.
• Leaving the workforce out. The people doing the job know which controls get quietly ignored and why. Ask them before you write the plan, not after.

Where to take this next

If you are the person responsible for getting this right in your business, the two courses that cover it properly are IOSH Managing Safely, which is the standard for line managers and supervisors, and the NEBOSH National General Certificate, which goes considerably deeper and is what most health and safety roles ask for.

If you would rather have somebody look at your controls with you, that is what our health and safety consultancy does: a visit, an honest gap list in priority order, and help working through it.