The Safer Choice

Chapter 1: Identify the hazards - the theory

1.1 Here's the theory

In the 2011 version of INDG 163 the HSE defined a hazard as “anything that may cause harm” giving examples of chemicals, electricity, working from ladders, or an open drawer. The 2014 version of INDG 163 dropped the definition, and as of 2021 there is no definition on the HSE website. However, well-known safety courses continue to use definitions based on previous HSE definitions with three elements:

  • A thing expressed as anything or something;
  • A possibility or probability expressed as “may” or “potential”;
  • Harm, damage or adverse effect, assumed to be to the health or safety of people but could be harm to the environment.

In the 2014 version of INDG, the first two of the five steps were:

    1. Identify the hazards
    2. Who might be harmed?
Not all hazards are this easy to identify

Image: Steve Buissinne at Pixabay

The original 1998 version of INDG 168 suggested that the second step included not just the who, but how they might be harmed. In the 2019 version, these two steps were combined into the single instruction to identify the hazards, but within this step to consider all those who might be harmed. As such, the person harmed is now an integral part of the definition of the hazard.

The law requires you to identify all reasonably foreseeable, non-trivial hazards. As the judge in a 1932 civil case put it, there is no need to protect against “fantastic possibilities” such as Mr Harcourt-Rivington’s normally docile terrier smashing the window of a car, causing an eye injury to Mr Fardon, who happened to be walking past at the time.

This finding has been reinforced repeatedly, including in R v Chargot (2008) “employers do not need to consider risks that are trivial or fanciful”. Even where prevention would be straightforward, if the harm were not foreseeable, an employer could not be expected to protect against it.

1.2 Why the theory doesn't work in practice

Legal cases can hinge on whether or not something should have been identified as a hazard. As the definition of a hazard given above suggests, we need to know what things are, how probable an adverse event is relating to that thing, and how harmful the thing could be. But we don’t know what Things are, we are not good at probability, and we leave people out during this step, and during all the other stages of risk assessment. We will look at each of these problems in turn.

1.2.1 We don't know what Things are

At school I learnt a rhyme called “The Nine articles of grammar”. It included the line:

“A noun’s the name of anything, as school or garden, hoop or string” (John Neale, 1886).

Hazards are easiest to identify when they are solid objects, such as hoops or string, or in the workplace, vehicles, machinery or tools. Although the HSE gives examples of activities (working from ladders) and invisible things (noise) “anything” or “something” draws us to more concrete objects. We see the ladder, rather than the work on the ladder.

A news story in 2019 showed how the “thing” that is the hazard can be misidentified. The headline accused the regulator of being “Daft as a Brush” with the sneering explanation that a furniture maker had been “banned from sweeping factory with broom after inspectors brand it a ‘health hazard’”. The article declared: “Brooms a health hazard? Quick, alert the families who have used them for generations.” The broom is the most visible thing in the photos of the furniture workshop featured in the article. But consider the case again:

Ask yourself: What is the thing with the potential to cause harm? What is the hazard?

Hopefully, you already realised that while the broom’s hazardous potential is sufficiently low as to be trivial, unless it is stored badly and tripped over, or if used in a way that created muscle strains. Or if it is used as a weapon to hit journalists who write bad articles. The non-trivial hazard is the wood dust, and one that the journalist and the business owner didn’t understand.
Even if you can identify all the physical things, even small grains of dust, that might be hazards, you might still miss something important. In the rhyme “Things” include places such as schools and gardens, or in the workplace, confined spaces, walk-in ovens, high roofs and cold rooms. The rhyme doesn’t mention that in some circumstances, actions can also be things. Entering a confined space, working at height, using a vibrating tool can all be hazards.

Is the lack of something really a thing? I frequently see risk assessments where “missing PPE”, “lack of training” or “no guard on machine” is stated as a hazard. But if goggles are needed to protect the eyes from flying debris, the debris is the hazard. “Not wearing safety goggles”, “lack of guarding” and “inadequate training” are failed controls.
Worse that this, I often see positive activities in the hazard column. The implication is presumably that the lack of the activity is a hazard, but this is not stated. For example, I’ve seen “housekeeping”, “training” and “communication” listed in the hazard column of risk assessments. One example on the HSE website includes “lighting” and “hygiene/comfort” in the hazard column.
Unless you are risk assessing the work of housekeeping, it is not the hazard. Unless you are risk assessing training activities (such as climbing a ladder during a work at height course) training is not a hazard. These are, at best, types of controls (which we’ll talk about in Chapter 3).
Another very common error in risk assessments is for the outcome of a hazard to be listed rather than the hazard itself. Common examples include:

  • slips and trips,
  • respiratory symptoms,
  • health of workers,
  • injury,
  • hand arm vibration syndrome (HAVS),
  • welfare and wellbeing.

What you can see in this list is a mix of negative and positive outcomes, or in the case of slips and trips, an intermediate stage between a hazard (such as a puddle of oil) and an outcome (such as a dislocated shoulder). These phrases are so often used together, that rather like “health and safety” the term loses its meaning. If you do not define the hazard clearly, your risk assessment will be less reliable.

The HSE suggests “A good starting point is to walk around your workplace and think about any hazards.” This “I spy” approach will inevitably miss some hazards, such as those created by routine work, maintenance or emergency procedures.

1.2.2 Probability is difficult

To be considered as reasonably foreseeable, the potential of a hazard for harm must be sufficiently severe, and the likelihood sufficiently high. This means that to decide if something should be identified as a hazard in step 1 you have to carry out step 2, asess the risks. Is my hot mug of coffee a hazard? The probability that the coffee leads to an accident at home is low, and the consequences would be limited. With more mugs of coffee, longer carrying distances, and more people around, the probability of a spill might be higher in an office, but still with limited consequences. If the coffee is carried from the teachers’ staff room to the infants classroom, the possibility of scalding a small child running in a corridor is both more likely, and with a potentially higher consequence.  

Whether the potential is high enough for the thing to be considered as a hazard is necessarily subjective. The well-illuminated step has such a low potential for leading to harm under normal use that it can be considered trivial. Under inappropriate use the potential could increase to a non-trivial point.

When an accident has never happened, it is easy to assume that it won’t. But, when an accident occurs, those who have suffered harm are likely to consider that the potential was significant. If the injured party takes a civil case against an employer for not identifying a hazard, or if the regulator determines that the hazard could have been identified, it is up to the courts to decide, with the benefit of hindsight, whether or not the thing should have been identified as a hazard.

When a three-year-old child died in hospital, having fallen down some steps in the playground, the HSE investigated (R v Porter, 2008). The child didn’t die from the fall, he died from a hospital-acquired infection. He would have recovered from the head injury he suffered in the fall. The HSE took two years to decide to prosecute the headteacher, and it took a further two years before the case came to court.

Steps might be a hazard. There is the potential for someone to fall down them and be hurt. Steps in a dark corridor, or in a home for confused blind people are more obviously a hazard. But for steps in a children’s playground, where there had been no injuries reported in nearly 30 years of children playing on the steps, does the probability reach the threshold to be defined as a hazard? For a crown court jury of lay people, the answer was yes. They found the headteacher guilty by a majority of eleven to one.

The little boy had not fallen, he’d jumped. It’s understood he was pretending to be Batman. The height he jumped was described in a news report (BBC 2008) as “that of a standard household settee.” He was nearer to four-years-old than three. Would you supervise a nearly four-year-old constantly in your own living room for fear that your sofa was “a hazard”?

Nine months after the jury trial, a panel of three judges overturned the jury decision. The judges understood that the tragedy of this child’s death did not alter the facts: the steps, under normal circumstances, were not a hazard of sufficient potential to warrant controls over and above normal playground supervision.

The “conkers bonkers” type stories illustrate the problem of assessing whether a hazard is trivial. Such stories are not always as straightforward as they might seem. Conkers might have become a significant and avoidable cause of playground accidents. In the 2019 “Daft as a Brush” story the journalists failed to appreciate that while sweeping the floor might be a trivial hazard when done at home, it becomes a significant hazard in a workshop full of an asthmagen like wood dust. HSE stats suggest that “carpenters and joiners are four times more likely to get (asthma) compared with other UK workers.”

This misunderstanding not only by the journalists, who might not be expected to know any better, but by the owner of the furniture company illustrates another problem that makes our assessment of the “potential” of something for harm more difficult. We are heavily influenced in our assessment of the likelihood of a harmful outcome by the timescales over which the outcome is realised. An injury from a fall from height is immediate and hence seems more likely than asthma or cancer from exposure to wood-dust, mesothelioma from exposure to asbestos, or long-term backache from slouching at our desks.

1.2.3 You have to consider people at every step

“Decide who could be harmed” should never have been thought of as a discrete step, after you have identified all the hazards. In practice you will have been considering “who could be harmed” when you are identifying the hazards. Or at least, you should have, which leads us to the third problem.

Some risk assessments suggest that the assessor is unable to distinguish between “the hazard” and the “person who could be harmed.” A well-known health and safety course uses an example of someone learning to drive to introduce the idea of risk assessment. The tutor’s manual gives the instruction “Explain to delegates that the cyclist is a hazard.”

Ask yourself: Who will be harmed if the car hits the cyclist? And which is the thing doing the harm?

If this high-profile course gets it wrong, is it any wonder that people producing risk assessments end up with the people who can be harmed in the hazard column? I’ve come across risk assessments with “pregnancy” or “pregnant worker”, “young person”, “disabled person” and even “customers” as the hazard. In the case of a customer, they might indeed be the hazard, but even then I’d want the risk assessment to spell out in what circumstances they could be the cause of harm. For any of the vulnerable categories, these are usually best considered in the “who can be harmed” column.

The primary legislation for health and safety in the UK explains which people should be considered in a risk assessment. With apologies for the non-gender neutral language of the law, the Health and Safety at Work etc Act (1974) requires that an employer should “conduct his undertaking in such a way as to ensure, so far as is reasonably practicable, that persons not in his employment who may be affected thereby are not thereby exposed to risks to their health or safety.” That is, by definition if your undertaking affects someone, you have a duty of care to not expose that person to health and safety risks (so far as is reasonably practicable – a concept we will look at in more detail later).

In the case of Donaghue v Stevenson (1932), a drinks manufacturer (Stevenson) tried to claim it had no duty of care to Ms Donaghue when she was made ill by the ginger beer it had bottled, because her friend had paid for it. The House of Lords however held that the manufacturer did owe her a duty of care, with the generalised finding that if it is reasonably foreseeable that someone will be affected by your acts or omissions, you owe them a duty of care.

The extent of this duty of care has been tested in many areas of law, as well as health and safety. In Caparo Industries v Dickman (1990) Caparo decided to buy shares in a company on the basis of an audit report that accountants Dickman had prepared for that company. The report had not been accurate about the profits of the company, causing Caparo to make a loss. However, the House of Lords determined that if anyone reading the report could claim for damages, this would open the floodgates to further claims.

The Caparo judgement was soon tested by those who counted themselves as victims of the Hillsborough disaster that resulted in nearly 100 deaths and over 750 injuries. The claimants were not at the ground, but they watched the events unfold on television, and claimed for the psychological harm they suffered. In Copoc, Alcock and others v. Wright (Chief Constable of the South Yorkshire Police) (1991) the appeal judges ruled that although it was foreseeable that those watching the match at home – knowing their loved ones were in the crowds – would suffer psychological distress, it would not be reasonable to extend the duty of care this far.

However, these two cases should not be taken to mean that duty of care is limited to those who suffer physical harm. In Walker v Northumberland County Council (1994) the judge ruled that an employer’s duty of care includes providing a safe system of work, and that includes work that does not cause undue stress. A social services officer had taken sickness absence for a stress-related illness. On returning to work his employer had agree to reduce his workload. Instead, on return to work his stress increased, as he tried to catch up on the backlog created during his absence. He was forced to retire early. In this case, the harm was measurable, and the first period of illness made it reasonably foreseeable that he would suffer illness again, even if other employees could cope with the same workload.

Many of the situations where people are not considered in risk assessments are more straightforward. I reviewed the risk assessments that a facilities company was providing to all its clients (overly generic risk assessments, but we’ll get to that later). The risk assessments of this facilities company had been accepted by multiple clients for years (which says something about their review process). The risk assessment for their staff using a vacuum cleaner on the stairs highlighted the electrical hazard of the machine, and the work at height hazard for the cleaning staff. What was completely omitted from the risk assessment was the trip hazard to the client staff created by the cable trailing up or down the stairs from the nearest socket. Some cleaning was done out of hours, so this would be a reasonable control for the hazard, but as in some buildings staff could be present at any time, this was a control that needed to be documented to protect client staff. Having failed to identify the client staff as those who could be harmed, the control was also omitted. Which leads us to the next problem.

1.3 How can it be better?

For something to be a hazard, it must be possible for someone to be harmed. So you need to evaluate the risk to a selection of representative people. Which means considering the the likelihood, the nature and the extent of that harm and how it is controlled. If you dismiss a hazard as trivial, because of its impact on most people, but haven’t considered its impact on vulnerable people, you have not identified hazards.
The trivial risk of a hot drink in an office might become significant in a school if small children could be exposed. Similarly, an otherwise trivial manual handling task could be harmful to someone with a pre-existing injury. If you don’t start to assess the risk for each type of person, you can’t decide if there is a non-trivial hazard.
But if it’s not a discrete step, how can we get started?
This ‘I spy’ approach recommended by the HSE will help you to identify obvious hazards, such as large, physical objects, or substances in containers with warning symbols. However, I spy overemphasises some hazards, and misses others, such as those created by routine work, maintenance or emergency procedures. The approaches in the next three chapters will help you identify hazards more effectively:

Chapter 2: Better identification through better understanding of what you are assessing
Chapter 3: Challenges to the norm
Chapter 4: How to include all the right people