NEBOSH National Diploma in

Occupational Health & Safety

UNIT NDA
MANAGING HEALTH AND SAFETY

element A2: LOSS CAUSATION AND INCIDENT INVESTIGATION

Sample material
(Material correct Autumn 2013)

RRC
27-37 St Georges Road London SW19 4DS United Kingdom
T +44 (0)20 8944 3100 F +44 (0)20 8944 7099
[email protected] www.rrc.co.uk Skype ID rrctraining
 L
Element A2: Loss Causation and Incident Investigation

A
Theories of Loss Causation

RI
Key Information

Incident studies have demonstrated that in any organisation there is a relationship between the number of

TE
major incidents and those with less serious outcomes.
The Single Cause Domino Theory suggests that in an accident there is a sequence of events or circumstances
that precede the harm, i.e.
Ancestry (i.e. upbringing).
Fault.
Unsafe act.

MA
Accident.
Injury.
Multi-causal theories suggest that preceding an incident there is a combination of causal factors at each level
that may combine to lead to the loss event.
Reasons model of organisational accidents states that for a major accident to occur a series of defences must
be defeated for the hazard to lead to a loss event. Unsafe acts may cause the failure of the defences. Unsafe acts
are made more likely by local conditions in the workplace.
LE
MP

Birds Accident Ratio Triangle

Other researchers have produced similar accident ratio

triangles:
Accident/Incident Ratio Studies
SA

There is no shortage of data on incidents such as

accidents or near misses. Some researchers have studied
the figures in detail and concluded that there appears to
be a relationship between the numbers of different types
of accident.
F. E. Bird used accident data to produce the following
accident triangle:
C
RR

Labour Force Survey 1990

RRC Training Unit A Element A2 | 2-1

 L
Element A2: Loss Causation and Incident Investigation

A
Domino and Multi-Causality Theories
One of the duties of the safety practitioner is to keep
details of accidents and ill-health conditions and carry

RI
out investigations. The law requires certain accidents
and occupational diseases to be reported. Often the
information that is recorded at the time of an accident
is not adequate for the purpose of investigation into the
cause, and so is certainly inadequate for the purpose of
preventing the accident happening again.

TE
For example, the report form may ask for the nature and
cause of the injury. This could be written as:
Reporting of Injuries, Diseases and Dangerous Occurrences Regulations Nature of injury - cut finger.
1995 (RIDDOR) Classifications
Cause of injury - caught on a sharp piece of metal.
The safety practitioner needs to know a lot more than

MA
this such as:
Which finger?
How serious was the cut?
Was this part of the normal job?
Should it have been sharp?
Should it have been there?
How should it have been handled?
Heinrichs Accident Triangle
LE
A good starting point in investigations is to consider the
The actual figures vary between the different accident two basic theories for accident causation.
triangles but the important thing to note is that for
every major incident or fatality, there are many more less Single Cause Domino Theory
serious or near-miss incidents. According to Heinrich:
Analysis also shows that: "A preventable accident is one of five factors in
MP

a sequence that results in an injury. The injury is

It is invariably a matter of chance whether a given
invariably caused by an accident and the accident
event results in injury, damage or a near-miss, i.e.
in turn is always the result of the factor that
near misses could so easily become more serious
immediately precedes it."
incidents.
Near-miss/less serious incident data can, therefore,
be a useful predictor of accident potential.
All events are due to failure to control so we can
SA

learn from even minor incidents.

The data from these triangles has a number of
limitations that you need to think about before trying to
apply it:
Not every near-miss or minor incident involves risks
which could actually have led to a serious incident or
fatality.
C

Be careful comparing:
Different triangles.
Different definitions (e.g. lost-time accidents). The five factors in Heinrichs accident sequence are
Different industries (with different types of risk). summarised in the following table.
RR

Statistical significance you need a certain amount

of representative data for a meaningful comparison
between your workplace and industry as a whole.

Unit A Element A2 | 2-2 RRC Training

 L
Element A2: Loss Causation and Incident Investigation

A
Heinrich's Accident Sequence of the accident sequence, perhaps the easiest factor
to eliminate is Number 3, the "unsafe act and/or
Accident Factors Description
mechanical or physical hazard".
1. Ancestry and Recklessness, stubbornness,

RI
social environment greed and other undesirable
traits of character that may
be passed along through
inheritance. Environment may
develop undesirable traits of
character or may interfere with

TE
education. Both inheritance and
environment may cause faults of
person.
2. Fault of person Inherited or acquired faults of
person such as recklessness,
violent temper, nervousness,
excitability. These constitute

MA
reasons for committing unsafe
acts or for the existence of
mechanical or physical hazards.
3. Unsafe act and/ Unsafe performance of persons
or mechanical or such as: standing under danger
physical hazard areas, careless starting of
machines, removal of safeguards
and horseplay; mechanical
or physical hazards such as
LE
unguarded gears or points of
operation, insufficient light,
which result in accidents.
4. Accident Events such as falls of persons,
striking of persons by flying
objects, etc. are typical accidents
which cause injury.
MP

5. Injury Fractures, lacerations, etc. are

injuries which result directly
from accidents.
The major point that Heinrich makes is that a
preventable injury is the natural culmination of a series
of events or circumstances which occur in a fixed logical
order. Here an analogy can be made with a row of
SA

dominoes placed on end such that if one falls it will cause

the next to fall and so on throughout the series (see
figure that follows). If one of the dominoes is removed
then the chain of events will be halted. In the same way,
consider Heinrichs accident sequence:

1. Ancestry and social environment.

2. Fault of person.
C

3. Unsafe act and/or mechanical or physical hazard.

4. Accident.

5. Injury.
RR

If this sequence is interrupted by the elimination of

even one of these factors, then the injury cannot occur
and the accident has been prevented. In the case

RRC Training Unit A Element A2 | 2-3

 L
Element A2: Loss Causation and Incident Investigation

A
RI
TE
Bird and Loftus extended Heinrich's theory to take into
account the influence of management in the cause and
effect of accidents, suggesting a modified sequence of
events:
MA
Heinrichs Domino Sequence

tend to be of a minor nature. A major disaster

normally has multiple causes, with chains of events,
and combinations of events. Fortunately, they are rare
LE
1. Lack of control by management. occurrences.
2. This permits the existence of basic causes (i.e. per- Systems Theory
sonal and job factors).
This is another way of looking at a multiple cause
3. In turn, this leads to immediate causes (such as sub- situation.
standard practices, conditions or errors). Factories and processes can be viewed as systems, i.e. an
MP

4. These are subsequently the direct causes of the ac- assembly of parts or components connected together
cident. in an organised way to perform a task, with inputs and
outputs and various kinds of control mechanisms.
5. Finally, this results in loss (which may be categorised A systems approach is often useful in simplifying complex
as negligible, minor, serious or catastrophic). operations. Part of the system can be taken as a 'black
This modified sequence can be applied to every accident box', with only the inputs and outputs considered.
and is of basic importance to loss control management. System failures are prevented or minimised by
components which cannot fail, by backup systems, or
SA

Multi-Causal Theories by redundancy built into the system (see Element A4).
There may be more than one cause of an accident, not Accidents happen in our system because it includes
only in sequence, but occurring at the same time. For fallible components such as machines and human beings.
example, a methane explosion requires: The system is operating in the failure mode.
Methane in the explosive range of 5% to 15%. You can see the essential features of the multiple
Oxygen, or air. causation approach in the following figure.

Ignition source.
C

The ignition will only happen if these three events occur

together. Each of the three events may, in themselves,
be the end result of a number of different sequences
of events. In accident investigation, all causes must be
RR

identified.
Usually simple accidents have a single cause, which is why
such events so frequently occur; but the consequences

Unit A Element A2 | 2-4 RRC Training

 L
Element A2: Loss Causation and Incident Investigation

A
RI
TE
MA
LE
Features of the Multiple Causation Approach
Immediate, Underlying and Root Causes might be considered unsafe and could therefore
There are various ways of classifying accident causes. foreseeably lead to an accident if not dealt with.
Remember that the same term may be used by different Note that an unsafe act or unsafe condition could alone
people to mean different things you can check this for result in an accident. For example, messing around is
MP

yourself by doing an Internet search on the above three an unsafe act which could take place in otherwise safe
terms. For consistency, we will use the terminology in conditions, but could nevertheless result in an accident.
the HSE guidance, Investigating Accidents and Incidents, Similarly, a person could be working in a perfectly safe
HSG245: manner, using safe equipment and materials, but suffer
Immediate cause refers to the direct cause of the injuries as the result of the collapse of a floor riddled
accident, i.e. the actual agent of injury or damage, with woodworm and dry rot. (You could argue, however,
such as the sharp blade of the machine. that collapse of the floor was due to an unsafe act, i.e.
failure to inspect the floor and supporting joists and to
SA

Underlying causes are the less obvious reasons for calculate the floor loadings.)
the incident - the unsafe acts and unsafe conditions,
such as the guard being removed. According to the accident sequence we discussed earlier,
unsafe acts and conditions are caused only by faults of
Root causes are the ultimate failings from which persons, and these faults are created by the environment
all other failings arise - typically management and or are acquired by inheritance.
organisational failings such as failure to train people
properly or failure to assess risks. The faults themselves generally arise because of
inappropriate attitudes, lack of knowledge or skill, or
We will now look at unsafe acts and conditions in more physical unsuitability.
C

detail.
An unsafe act is human performance which is contrary
to accepted safe practice and which may, of course,
lead to an accident. Unsafe conditions are basically
RR

everything else that is unsafe after you take away unsafe

acts. So, this is the physical condition of the workplace,
work equipment, the working environment, etc. which

RRC Training Unit A Element A2 | 2-5

 L
Element A2: Loss Causation and Incident Investigation

A
Reasons Model of Accident Causation
Topic Focus

RI
Topic Focus
Active failures are one cause for the barriers to
Latent and Active Failures be defeated.
Rather than using the words immediate, Active failures are those unsafe acts which have
underlying or root causes, the terms latent immediate effects on the integrity of the system

TE
and active failures are also commonly used. and are usually committed by those directly
involved in the task. Such individuals often
Following research into a series of disasters,
suffer directly as a result of the incident and
James Reason (an occupational psychologist)
may often be blamed as well. The cause of the
has developed a model of accident causation
failure will be due to an error (accidental) or a
for organisational accidents. An organisational
violation (deliberate). Such unsafe acts are made
accident is rare, but if it happens often has
regularly but few will cause the defences to be
disastrous consequences (e.g. Piper Alpha,

MA
penetrated, an example being the chemical plant
1988). Reasons model shows that organisational
operator who opens a valve allowing a hazardous
accidents do not arise from a single cause but
substance to escape.
from a combination of active and latent failures.
The model then shows that the local workplace
factors influence the chance of an unsafe
act occurring. In the case of the hazardous
substance escape, this may be due to a lack of
supervision or training, maintenance failure,
unworkable procedures, etc.
LE
According to the model the local workplace
factors are affected by decisions made at
a strategic level by senior management,
government, regulators, manufacturers, etc.
In the case of senior management this might
be lack of recognition of the importance of
occupational health and safety, which will be
MP

reflected in the culture of the organisation by

the behaviour that is considered acceptable.
The management may give safety a low priority
with no commitment and minimal funding.
These failures at the strategic levels, both in the
organisation and the external environment, are
described as latent failures because they remain
dormant and possibly unrecognised until they
SA

Adapted Version of Reasons Model of Accident Causation interact with the local factors and the unsafe
acts and work environments and increase the
In the model there is a series of defence barriers
likelihood of an active failure.
between the hazard and a major incident. These
not only prevent the incident, (e.g. containment When the gaps created by active failures align
of the hazard, safe operating procedures, with those created by the latent conditions, the
etc.) but also provide warning of danger (e.g. opportunity exists for a serious outcome.
an alarm) and mitigate the consequences
(e.g. means of escape). These multiple layers
characterise complex technological systems such
C

as a chemical plant.
However the barriers are not perfect and can be
defeated.
RR

(Continued)

Unit A Element A2 | 2-6 RRC Training

 L
Element A2: Loss Causation and Incident Investigation

A
Categories of Unsafe Acts
Unsafe acts of persons may be categorised under the
following headings:

RI
Operating without authority.
Operating or working at an unsafe speed.
Making safety devices inoperative.
Using unsafe equipment, or using equipment

TE
unsafely.
Unsafe methods, e.g. loading, carrying, mixing.
Adopting an unsafe position or posture.
Working on moving or dangerous equipment.
Messing/playing around, e.g. distracting, teasing,

MA
startling.
Failure to wear safe clothing or personal protective
devices.
Lack of concentration; fatigue or ill-health.
From this list you can see that unsafe acts may either
be deliberate violations (sometimes called 'active') or
unintentional errors (sometimes called 'passive'). We
discuss these human factors in detail in Element A7.
LE
Categories of Unsafe Conditions
The following categories describe unsafe conditions
from which an accident may result:
Inadequate guarding; guards of inadequate height,
strength, mesh, etc.
Unguarded machinery, or the absence of the required
MP

guards.
Defective, rough, sharp, slippery, decayed, cracked
surfaces.
Unsafely designed machines, tools.
Unsafe arrangements, poor housekeeping,
congestion, blocked exits.
Revision Questions
SA

Inadequate lighting, glare, reflection.

Inadequate ventilation, contaminated air.
1. Outline the five factors in Heinrichs accident
Unsafe clothing - no goggles, gloves or mask.
sequence.
Unsafe processes - mechanical, chemical, electrical,
nuclear. 2. How does Bird and Loftus theory of accident
causation differ from Heinrichs?
Hot, humid or noisy environment.
3. According to Reason, what in an organisation
C

are latent failures?

4. What important principle of accident

causation theory do accident ratio studies
illustrate?
RR

(Suggested Answers are at the end of Unit A.)

RRC Training Unit A Element A2 | 2-7

 L
Element A2: Loss Causation and Incident Investigation

A
Quantitative Analysis of Accident and Ill-Health Data

RI
Key Information

The amount of injury and ill-health in a population may be described by calculating the accident/incident

TE
frequency rate, the accident incidence rate, the accident severity rate or the ill-health prevalence rate.
Bar charts, pie charts and line diagrams can be used to represent incident data in a graphical format.
Statistical variation within a population may be described using a normal distribution.

Calculating Loss Rates from Raw Data

MA
Jargon Buster Topic Focus

Incidence Accident Incidence Rate

This is calculated from:
Incidence reflects the number of new cases of
a particular event in a population over a given Number of work-related injuries 1,000
time (e.g. a year) and is often used to describe Average number of persons employed
LE
accidents as each accident is a new event.
It is a measure of the number of injuries per
Prevalence 1,000 employees measured over a defined
Prevalence is the total number of cases in a period, e.g. a year.
particular population as a proportion of the total
population. It is often used to represent ill-health Accident Severity Rate
statistics and reflects not only new cases but also This is:
MP

those who continue to suffer.

Total number of days lost 1,000
Total number of man-hours worked
It is a measure of the average number of days
lost per 1,000 hours worked and gives the
Topic Focus average number of days lost per accident.

In making comparisons between various Ill-Health Prevalence Rate

SA

industries, or between work areas in the same Prevalence is a term often used to describe
factory, it is useful to consider the commonly ill-health in terms of the proportion of persons
used injury ratios. who have the prescribed ill-health condition at a
particular time. The rate is calculated as:
Accident Frequency Rate
This can be calculated from: The total number of cases of ill-health
in the population 100
Number of work-related injuries 100,000 The number of persons at risk
Total number of man-hours worked The calculation gives the percentage of the
C

It is a measure of the number of accidents per population with the disease.

100,000 hours worked.
(Continued)
RR

Unit A Element A2 | 2-8 RRC Training

 L
Element A2: Loss Causation and Incident Investigation

A
Statistical and Epidemiological Analyses may show the trend better than others. It is also possible
to include a measure of severity, e.g. by plotting the
in the Identification of Patterns and number of days lost through sickness, or the costs of
Trends damage/repair.

RI
Epidemiological analysis may identify a pattern in data
distribution but it does not in itself give information on
why the pattern is occurring. The pattern must then
be analysed to determine whether causal factors can be
identified and remedial action taken. Epidemiology is

TE
used to identify problems which would not be apparent
from single incidents, e.g. to establish whether a number
of individual cases of food poisoning are linked and,
therefore, may constitute an outbreak.
It is only possible to carry out an epidemiological analysis
when the same type of information is available for all of
the accidents being analysed. Typical data dimensions

MA
include location of accident, time of accident or nature
of injury, etc. Single dimension analysis looks at just
one dimension, e.g. location of accident/incident. The
analysis would involve looking for deviations from what
would be reasonably expected. For example, if work was
spread evenly over a number of sites (with all sites being
of comparable size and carrying out similar jobs) then
you would expect that the numbers of accidents at each
site would be evenly spread. Peaks and troughs should
LE
be investigated.
Statistics is concerned with systematically collecting,
organising and interpreting numerical data. You can see that the different types of analysis may
Epidemiology looks at occurrences of disease in different identify different patterns or trends which are useful
groups of people and tries to identify a cause and in identifying what is happening within an organisation
prevention/control strategy. The size of groups looked and can help to identify what actions must be taken to
at can be quite small, e.g. comparing the incidence of a improve safety performance.
disease between two factories, or very large, in which
MP

comparisons are made between the populations of Presenting and Interpreting Loss Event
different countries. Epidemiology is so-called because
it was initially concerned with the study of epidemics
Data
(a widespread outbreak of an infectious disease). Both We will consider some typical ways in which data can be
types of analysis are useful in identifying patterns and presented in this section.
trends and giving an insight into any necessary remedial
action that may be required to overcome a particular Histograms
problem. 'Histogram' is the name given to a particular type of bar
SA

chart. It is the diagram used to illustrate a frequency

Accident and incident data can be used to measure
distribution and it always has the following features:
whether performance is improving or deteriorating and
to compare data over time, e.g. this year's figures with All the columns touch each other.
last year's. This trend analysis, however, can be affected Both axes have scales:
by a number of factors other than whether the safety
management systems in place are effective. For example, The horizontal axis carries the variable under
an increase in the amount of work carried out by an consideration.
organisation may lead to more accidents but this does The vertical axis shows the frequency with which
C

not necessarily mean that the safety performance has the values of the variable occur.
deteriorated. In the same way, a reduction in work may The bars are all the same width but the values of the
lead to a reduction in accidents whether or not there are variable need not begin at zero, i.e. the first column
any changes to the safety management practices. of the histogram need not touch the frequency axis.
RR

The simplest method of trend analysis is to plot on a

graph the number of accidents or incidents against
time. The measure of time used, e.g. monthly, quarterly,
annually, should be considered, as different time periods

RRC Training Unit A Element A2 | 2-9

 L
Element A2: Loss Causation and Incident Investigation

A
Example 1
The following table shows the simple frequency
distribution of the number of days' absence caused
by 60 lost-time accidents. (Columns 1 and 2 form the

RI
frequency distribution, columns 3 and 4 are calculated as
shown from the frequency distribution.) The histogram
(the next figure) is based on this table.

Days Lost per Accident

TE
No. of Accidents Man-Days Cumulative
Days lost per Accident Causing Lost-Time Lost Frequency
x f fx fcum
0 3 0 3
1 5 5 8
2 7 14 15

MA
3 12 36 27
4 9 36 36
5 8 40 44
6 6 36 50
7 5 35 55
8 1 8 56
9 3 27 59
10 1 10 60
LE
Totals 60 247 -
MP
SA
C
RR

Lost-Time Accidents Causing Specified No. of Days Absence

(Note: The mode is the most popular frequency, i.e. 3
days.)

Unit A Element A2 | 2-10 RRC Training