Incidents in manual handling activities - UQAM

ELSEVIER Safety Science 21 (1996) 223-237
Incidents in manual handling activities
M. Lortie a* * , R. Pelletier b
a SC. Biologiques, UQAM, CP 8888, Succ. Cenfre-ville, Montr&l, Qc, CunadaH3C 3P8
b Associution Paritaire pour la Sante’ et la ShuitP du Truvail du Secteur de l’Hahillement, 9310 St-Laurent,
Mont&al, Qc. CannduHZN IN4
Abstract
More than 20 000 handling manoeuvres in loading or unloading trailers in a large transportation
company were observed. The aim of the study was to determine the frequency and nature of the
incidents occurring whilst handling. The incidents were recorded and interpreted from a double
perspective, namely as risk factors and as activity regulation factors. The study showed that on
average, one incident occurred in every seven handling manoeuvres: 7 1% of them were related to
the environment, and 29% to the activity. While numerous, the environment-related incidents were
generally without consequences. In this respect, the activity incidents appeared more risky; also,
the majority of these incidents occurred during the load positioning phase. The study also showed
that loads are re-handled twice as often in unloading as in loading, the frequency of re-handlings
varying widely from one handler to another. The reasons why so few of the observed incidents are
reported in accident studies and their significance in the understanding of handling problems are
discussed.
Keywords: Incident; Manual handling; Accident; Activity regulation; Transport industry
1. Introduction
Incidents have been considered for a long time as key elements in understanding
accidents. Abundant scientific literature associates accident occurrence with a dysfunc-
tion in the man-machine system, where the accident is the final stage in a modification
in the normal state of the system (Leplat, 198.5; Kjellen, 1984). Furthermore, factors
characterizing possible dysfunctions have been integrated into accident models, reveal-
ing the intrinsic importance of these variations in the state of the system (Leplat and
Rasmussen, 1984).
* Corresponding author.
09257535/96/$1.5.00 0 1996 Elsevier Science B.V. All rights reserved
SSDI 0925-7535(95)00066-6

224 M. Lortie, R. Pelletier / Safety Science 21 (1996) 223-237
Incidents are dealt with from two perspectives. In the first, incidents are part of a
functional flow diagram intended to express, logically and chronologically, the situations
and events that resulted in the accident. The cause tree developed by the INRS in France
is a typical example. The second type is more taxonomic in nature: it characterizes
incidents, and their significance in the origin of accidents is interpreted in relation to
their frequency and/or their rank in the sequence of events leading to the accident. The
MAIM (Merseyside Accident Information Model) system, proposed for studying han-
dling accidents, is one example (Troup et al., 1988). In both cases, the information is
collected following an accident involving injury.
The concept of incident can also vary greatly, depending on whether the studies focus
on risks or on system regulation. For example, Hoyos and Zimelong (1988) link
incidents to risky events, which are, therefore, defined in relation to their potential for
generating accidents, according to the following sequence: unsafe incident, incident with
material damage, potential accident, and accident with injury. With Leplat and Ras-
mussen (1984), however, an incident is part of a system-regulation process; the goal of
the operators’ actions is, therefore, to return a system that is not operating as expected to
the normal state. In this process, operators are reliability agents who must anticipate
incident recovery. Here, the concept of incident therefore has a broader scope because it
is defined without exclusive reference to the risk of injury. In this latter case, risks can
be anticipated not only from analysis of incidents, but also of the work activity itself.
Up until now, overexertion musculoskeletal injuries, particularly back injuries and
those associated with handling activities, have been the subject of numerous studies. The
notion of incident is conveyed mainly from two perspectives. In the first, whether a
dysfunction or an incident occurs or not is used to characterize the “fortuitous” nature
of the accident. This distinction was introduced by Shannon and Manning (1980) to
differentiate cases in which the injury is itself the only unanticipated event from those
cases in which the injury results from another unanticipated event (called a true
accident). With the second perspective - which coexists with the first - the incident
is analyzed mainly as an injury mechanism (e.g., loss of balance, impact); it is defined in
relation to the accident victim, without reference to the system or to any notion of
regulation. It should furthermore be emphasized that the handling activity itself has been
the subject of only a few studies; the injury mechanism is almost always the only point
of interest.
A study dealing with 611 accidents involving handlers loading and unloading trailers
on loading docks showed that one incident or unexpected event other than the accident
itself was reported in 58% of the accidents; furthermore, the incidents were of a more
varied nature than what is generally reported in the literature (e.g., loss of handgrip,
unexpected instability of the load; Lortie et al., 1996). In addition, interviews with the
handlers revealed that incident prevention was a deciding factor in the choice of method
(Authier and Lottie, 1993). Incidents that preoccupy handlers are, however, not neces-
sarily risky events. They may be events that occur in system regulation. For example,
handlers prefer to use, depending on the circumstances, a method that allows direct load
positioning, where a readjustment of the final position of the load is unnecessary.
Despite the apparent importance of this phenomenon, no study dealing with the
observation of incidents during handling could be found.

M. Lnrrie, R. Pelletier/Sufety Science 21 (1996) 223-237
The goal of this study was, therefore, to verify the importance of incidents in
handling activities, whether they are risky events or not.
2. Method
2.1. Subjects and activities observed
The manual handling activities of 31 handlers loading (n = 14) and unloading
(n = 17) trailers were filmed for one shift each. The subject of the handling could be an
object such as handling equipment (e.g., pushing on a cart, changing the forks on a
forklift).
Generally, unloading work consists of identifying the loads and grouping them into
lots by destination. Goods transferred manually are generally grouped on carts that are
then brought into the trailer zone corresponding to their destination. They are loaded into
the trailer by another handler, whose purpose is to fill the trailer as full as possible and
to make sure that the load is stable and balanced. These moved goods vary greatly in
type, weight, volume and format, as well as in their arrangement in the trailers or on the
dock. The activities are described in greater detail in Baril-Gingras and Lortie (1995).
Some 20000 handlings were filmed.
2.2. incident observation
We have defined an incident as being any unanticipated or undesired event. The
incident typology was developed from what was mentioned in the accident reports and
interviews. However, since data collection based on observation and verbalization is
consistent with two different reasonings, we had to structure incident collection in a
specific way right from the start. For example, some incidents reported as the first or
only event in accidents are observed better as the consequence of another incident. For
instance, “unusual exertion” results almost systematically from another event - it is,
therefore, seen mainly as a consequence - while it is often reported as the only event in
accidents. Furthermore, specific circumstances or incidents reported in the accident
declaration cannot necessarily be easily translated into “observables” (e.g., good is
heavier than expected or in a difficult position). Conversely, easily observable events
may be difficult to report in an accident context, particularly if they are seen as partly
related to individual skills. This is the case of “difficulty moving” or “difficulty
controlling” a load. They are, however, reported in a broader interview context.
The variables retained and the observation criteria are summarized in Table 1. The
first part of the grid characterized the circumstances of each handling. Then, the
observer verified whether there was an incident, and if so, whether the incident involved
the activity or the environment, the result of the incident, as well as the handling phase
in which the incident occurred (e.g, pick up, transfer, placement, during the entire
handling). In cases where an incident involved both the activity and the environment, the
observer determined the sequence.
225

226
Table 1
Observed variables
1. Handling circumstances
aObject moved:
.Foot support:
� Teamwork:
ONanne of handling:
2. Activity-related incidents
� Difficulty controlling:
� inaccurate positioning:
� Difficulty moving:
@Loss of balance:
@Sudden manoeuvre:
� Impact/catching on
something:
3. Enuironment-related incidents
@Instability:
@Access difficulty:
� Defect:
4. Consequences
� On the activity:
� On the environment:
5. Timing of the incident
M. Lortie, R. Pellefier/ Safety Science 21 (1996) 223-237
goods (box or other), equipment (a part or all), structural element
(e.g., plate), other.
same level (ground or pallet), one foot on the ground, climbed
on goods, climbed-other.
alone, teamwork (series, parallel, both together),
with the lifter.
regular (final pick up and placement) versus additional
(before or after regular, other). Regular handling manoeuvm:
executed all at once or in sequence (e.g., when climbing).
incident relating to grasp or to the displacement procedure chosen.
Typically the handler loses his handgrip, has diffkulty finding
the way to pick up or position the object.
goods need to be placed again because the placement is not as
accurate as anticipated.
the object is particularly difficult to handle due to its
weight/and or volume, even if the situation is under control.
with or without falling; it can be a slight loss of balance,
insofar as it is observable.
it can be to prevent an incident or to respond to a degrading situation.
between the handler and a fixed object.
any unanticipated movement of an object, piece of equipment,
or element in the environment. For example, a box falling,
a van moving, a cart moving.
situations where other objects hinder or particularly complicate pick
up or placement.
condition differing from the usual. For example, a handle that breaks,
a wheel that jams, a hole in the ground.
additional manoeuvre, sudden manoeuvre, overexertion, loss of balance,
impact, modification in the operative mode, other.
breakage. jamming, instability, other.
� While handling (pickup, transfer, placement, all the time), transporting, waiting, moving without a load.
The nature of the object moved, the level of foot support, whether teamwork was
involved, and the nature of the handling were character&d. In this last case, so-called
regular handling where the placement position is final, was differentiated from so-called
additional handling, which could take place prior to the final positioning (e.g., sorting
loads) or after the final positioning (e.g., repositioning a load). In reality, and contrary to
what is usually expected, the same object can be handled more than once. The
rehandling of goods is generally considered by handlers as something to be avoided.
An activity-related incident is an event observable in the activity; it may be an
obvious inability to control or move the object, a loss of balance, an unanticipated or

M. Lortie, R. Pelletier/ Sufety Science 21 (1996) 223-237 221
sudden handling, a collision with an immobile object, or an inaccurate handling
manoeuvre in which the object does not reach the desired final position. Whilst such
incidents have no pertinence in terms of risk, they were, however, retained as an
undesired event. This was identified during interviews with handlers as one of the
factors considered in choosing a handling method (Authier and Lortie, 1993). Environ-
ment-related incident is an event caused directly by some dysfunctionment in the
material environment; it may be an instability of goods or equipment, a specific access
difficulties or a defect of the material. It has to lead to an observable undesired event.
The observed consequences could be the initiation of another incident, namely an
undesirable a priori event such as a rehandling or a change in procedure.
This differentiation of the incidents in two groups was based on the need to facilitate
the observational procedure. As a fact, activity and environment are closely related: it
was important to avoid that the observer had to decide constantly if the origin of the
incident was in the environment or in the activity. Therefore, the differentiation retained
was more procedural than conceptual. The observation criteria are presented in Table 1.
All the handling manoeuvres were observed first by an observer, and incident identifica-
tion verified by a second observer.
3. Results
3.1. Main characteristics of handling activibes
The main characteristics are summarized in Table 2. More than 20000 handling
manoeuvres were observed, divided almost equally between loading and unloading. The
Table 2
Handline manoeuvres observed: their structure and main characteristics
Object moved
Merchandise:
regular handling manoeuvre
additional handling manoeuvre: pre
post
Handling equipment
Other
Level offhot support
Both feet on the ground or on pallet
One foot on the ground
Climbed on merchandise
Climbed, other
Other
Teamwork
Alone
In a team: serial or parallel work
both together
Loading % Unloading % Total %
(n=10134) (n=10539) (n=20673)
71.9 42.6 67.2
I .5 14.1 8.2
8.4 5.0 6.6
16.5 16.6 16.6
1 .I 1.1 1.4
92.9 96.7 94.8
I .7 0.5 1.1
4.4 1.4 2.9
0.9 1.2 I.1
0.1 0.2 0.1
87.9 95.1 91.6
9.0 3.6 6.2
3.1 1.3 2.2

228 M. Lortie, R. Pelletier/ Safety Science 21 (1996) 223-237
objects moved were goods and equipment in 82% and 17% of the cases, respectively.
When the goods were handled, 18% additional manoeuvres were noted, with the
phenomenon being twice as common in unloading (L: 12%, U: 24%). In loading, these
additional manoeuvres, in 85% of the cases, followed a regular handling manoeuvre:
most often, the handler decided to move the already loaded goods again. In unloading,
on the contrary, 75% of these additional handling manoeuvres occurred before the
regular handling. This corresponded mainly to locating or sorting operations. Also, 2%
of the regular handling manoeuvres were carried out in at least two separate steps (i.e.,
two picking up-transfer-placement cycles), with the phenomenon being twice as preva-
lent in unloading. These occurred mainly when climbing was involved. Only 8% of the
handling manoeuvres involved teamwork, organized 3 times out of 4 in series (one
handler passes the goods to the other) or in parallel (the two workers work with the
same cargo, but independently). Simultaneous lifting of the same goods occurred in only
2% of the handling manoeuvres. In 95% of the cases, the handler had both feet at the
same level, either on the ground or on a pallet. Climbing was observed twice as often in
loading as unloading (L: 7%, U: 3%).
3.1 .I. Inter-handler variations
Table 3 shows that on average, 33% more goods were handled when loading than
when unloading, and that the number of moved goods varied significantly from one
handler to the next and from one work shift to the next, particularly in unloading. The
difference observed between the minimum and maximum values reached a ratio of
almost 5 (752/158). The number of goods was, however, not a good indicator of the
workload; a small number of goods to be moved manually can represent either an easy
cargo where the forklift will be extensively used, or a difficult cargo consisting of large
objects. The number of additional manoeuvres is on the other hand a good indicator of
difficulty. As shown in Table 3, the proportion of loads re-handled varied from 6% to
77%.
Table 3
Handling manoeuvre frequency: inter-handler variations
Loading ( n = 14)
# of # of ad- % of ad-
loads a ditional ditional
manoeuvres manoeuvres
Mean 498 70 14.1 b
Standard deviation 128 35
Median 514 65
Maximum ’ 702 129 50.0
Minimum ’ 259 25 5.7
Unloading ( n = 17)
# of # of ad-
loads ditional
manoeuvres
376 122
135 58
361 91
752 144
158 43
% of ad-
ditional
manoeuvres
a Some handlers assigned to loading may go to help with unloading for short periods and vice versa. These
data were excluded.
b Individual average % were 11.8 for the loaders and 22.2 for the unloaders.
’ These are the absolute maximum and minimum values. Values on the line may refer to different handlers.
32.4
76.9
13.9

M. Lortie, R. Pellefier/Safety Science 21 (1996) 223-237 229
Table 4
Characteristics of the incidents occurring during loading and unloading of the trailer
Activity-related incident Loading % Unloading % Total %
(n = 462) (n=400) (n= 862)
Difficulty controlling 296
Inaccurate handling 37.7
Difficulty moving 17.5
Loss of balance 12.8
Sudden handling 1.5
Impact/catching on 0.7
Other 0.2
27.2
28.5
25.0 31.8
34.5 25.4
8.5 10.8
2.0 1.7
1.0 0.8
1.7 0.9
Load Merchandise box 63.8 55.3 59.8
other 28.0 33.3 30.4
Equipment 5.2 8.5 6.7
Other object 3.0 3.0 3.0
Timing Picking up 12.3 28.3 19.7
Environment-related incident
Transfer 11.5 10.7 .ll.l
Placement 58.4 32.5 46.4
All the time 8.9 13.7 1 I.1
Other (and other object) 8.9 14.7 11.6
Type Instability
Access difficulty
Defect
Load Merchandise
Equipment
Other object
Timing Picking up
Placement
Other
Unrelated
3.2. Incidents
(n= 902) (n= 1187) (n=2089)
54.2 63.7 59.6
34.2 29.7 31.7
11.6 6.6 8.7
box 41.2 40.2 40.6
other 21.0 8.2 13.7
34.3 48.5 42.4
3.5 3.1 3.3
7.4 2.9 4.9
9.2 3.5 6.0
0.6 0.8 0.7
82.8 92.8 88.5
On average, one incident was observed for every seven handling manoeuvres (L:
13,5%, U: 15,1%); seven times out of ten, it involved the environment (Env: 71%, Act:
29%). In 14% of the cases, there were two incidents, one involving the activity and one
involving the environment. The typology of these incidents is presented in Table 4.
3.2.1. Activity-related incidents
Three types of incidents explain 85% of the observed incidents: inaccurate position-
ing (32%), difficulty controlling the object (28%), and difficulty moving the object
(25%). Loss of balance ranked fourth with 11% of the incidents. Unplanned or sudden

Table 5
Timing of the incidents in relation to the handling activity
all the
time
Type of incident Loading
Unloading
Total
pick transfer place all the pick transfer place all the pick transfer place
UD % % ment% time % up % % ment % time % uo% % ment%
Control difficulty 24.8 18.6 54.8 3.9 45.1 17.6 30.4 6.9 33.8 18.2 42.9 5.2
Moving difficulty 10.7 16.0 25.3 48.0 33.0 10.4 16.5 40.0 24.2 12.6 20.0 43.2
Loss of balance 12.2 29.3 58.5 0.0 30.0 30.0 35.0 5.0 I8,O 29.5 50.8 1.6

hf. Lortie, R. Pelletier/Safety Science 21 (1996) 223-237 231
manoeuvres represented only 2% of the incidents involving the activity. These incidents
were, however, distributed differently depending on whether loading or unloading was
taking place. For example, “inaccurate positioning” and “loss of balance” were one
and one-half times more prevalent in loading, while conversely, “specific difficulty
moving” objects was observed twice as often in unloading.
Most incidents occurred in handling a load, most often a box (60%). However, boxes
made up more than 90% of the moved goods. Boxes were more often involved in
incidents during loading than unloading (L: 64%, U: 55%). A more detailed analysis of
the type of incident in relation to the type of goods showed, however, that goods other
than boxes were the source of 55% of incidents where difficulty moving the goods was
observed, and of 49% of cases involving loss of balance.
It was also noted that incidents generally occurred at a specific stage in the handling
manoeuvre, most frequently at placement (46%). The transfer itself most often occurred
without incident. However, it was noted that the handling step where most incidents
occurred differed, depending on whether loading or unloading was taking place. In
loading, incidents occurred significantly more often at placement (58%), while in
unloading, they were observed almost equally in picking up the load (28%) and at
placement (32%). The analysis was therefore expanded to relate the handling step to the
type of incident, except for “inaccurate positioning” which, by definition, essentially
refer to placement. The results in Table 5 show that “control difficulties” were more
prevalent during placement when loading (53%), and in picking up the load during
unloading (45%). Loss of balance was also more prevalent in placement during loading
(58%), while in unloading, it was observed almost equally at all stages in the handling.
Finally, “specific difficulty moving” was, in loading, twice as frequent in placement as
in picking up the load, while the opposite was true in unloading.
A comparison of the handling circumstances of these incidents (Table 6) to the
circumstances in all of the handling manoeuvres (Table 2) revealed that at the time of
these incidents, the handlers had climbed twice as often. The difference was especially
marked in loading (L: 18%, U: 7%). When the handler was climbed, losses of balance
were the most frequently ensuing incidents; however, they represented only one third of
Table 6
Circumstances related to the Occurrence of activity incidents
Foot support
Both feet on the ground or on pallet
One foot on the ground
Climbed on merchandise
Climbed, other
Other
Teumwork
Alone
In a team work serial or parallel
together
Loading Unloading Total
82.0 93.0 87.1
2.6 0.7 1.7
11 .I 2.5 7.4
3.2 3.5 3.4
0.4 0.3 0.3
88.9 93.0 90.8
6.1 1.3 3.8
5.0 5.7 5.3

232 M. Lortie, R. Pelletier/Sufety Science 21 (1996) 223-237
the observed incidents. Teamwork was also more prevalent with incidents, but the
difference was essentially due to situations in which the handlers move the same object
at the same time.
As mentioned in the observation grid description, incidents may initiate other
unanticipated activity-related or environment-related events; some of these consequences
are potentially risky, while others are not. The analysis shows that 50% of the incidents
led to consequences in the activity (L: 53%, U: 45%). However, in 72% of the cases, the
initiated event had no specific risk. Mainly, additional manoeuvres or changes in
methods were involved. Incidents during unloading nonetheless produced dangerous
consequences, as for example, loss of balance, overexertion (L: 22%, U: 34%).
However, only 5% of these incidents had an impact on the environment; basically, it
was instability in the goods or equipment.
Loss of balance was observed as an activity incident or as a result of another incident.
Particular attention was paid to this since it is often reported as a major accident
circumstance. In this study, some imbalance in the handler was observed on 93
occasions as a first event, and on 28 occasions as the result of another incident (Act:
71%, Env: 29%) in a total of more than 20000 handling manoeuvres. Only four of these
losses of balance resulted in a fall. Finally, 92% of these losses of balance, as a first
incident, occurred during the handling of goods, and 63% during loading. In 40% of the
cases, the handler had climbed, or was not working with both feet at the same level.
3.2.2. Environment-related incidents
Cases of instability dominated in loading and unloading (L: 54%, U: 64%). Access
difficulties represented approximately one third of the incidents (L: 34%, U: 30%) and
defects 9% (L: 12%, U: 7%). Contrary to activity-related incidents, the characteristics of
environment-related incidents were practically the same for the two functions. In this
type of incident, the object involved was often something other than the object being
handled. Thus, 42% of these incidents involved equipment, particularly carts (L: 34%,
U: 49%), while the equipment was being handled at the time of the incident in only 10%
of the cases. Furthermore, Table 4 shows that in 89% of the cases, the moment at which
the incident occurred was unrelated to the handling being carried out (e.g., a cart moved
while the handler was doing something else a few feet away).
The two main sources of instability were goods and equipment, but it was the latter,
particularly during unloading, that was most often involved (L: 59%, U: 72%). These
instabilities had no impact on the activity in 78% of the cases; the handler in fact
performed no recovery action during or after the incident (L: 79%, U: 77%). When
recovery took place, it was essentially to return the object involved to its original
position. Only 6% of these instabilities required a potentially risky response (L: 5%, U:
7%), the most prevalent being an unanticipated movement to control the instability.
The same picture emerges for all of the incidents: in 70% of the cases, the incident
had no impact on the handler’s activity; recovery had a potential risk in only 9% of the
cases. Only 4% of these incidents had an impact on the environment; a series of events
was therefore rare. Finally, similar to activity-related incidents, boxes were more often
involved than other goods (74% versus 25%), particularly in unloading (L: 66%, U:
83%).

4. Discussion
M. Lortie, R. Pelletier/Safety Science 21 (1996) 223-237 233
As mentioned in the introduction, the study of incidents falls within the scope of
studies whose purpose is sometimes the study of risks and sometimes the study of
system regulation. In this study, we have attempted to combine the two objectives.
Studies focusing specifically on each of these objectives would certainly have consid-
ered elements that are not dealt with here. Hence, a study focusing on regulation would
have included parameters characterizing the regulatory response. This was not done. A
study focusing on risks might have included an a priori evaluation of the “safety” of the
activities or the environment as such. Here, no attempt to characterize the environment
beforehand (e.g. the condition of the floors) was made. Incidents were noted only if they
resulted in observable events. However, considering the two objectives simultaneously
produces a rather more complete picture of the unanticipated or undesired events.
4.1. Incidents as risk factors
Accident analysis (Lortie et al., 1996) has shown that approximately one third of the
injuries were caused by an object other than the handled object; observation shows that
environment-related incidents are in fact prevalent. For another third of the accidents, it
was the moved object that injured the handler. Incident analysis confirms that incidents
such as “control difficulties” or “equipment defects” are prevalent. An incident was
reported in 58% of the accident declarations. The high rate of observed incidents in fact
reveals an activity where there are many imponderables, even if they are often without
consequence. The incidents most often reported in accidents-instability (25% of the
reported incidents), control difficulties (19%), and defects (17%) correspond to regularly
observed events. The picture emerging from incident observation and accident analysis
can therefore be said to be compatible.
Most incidents are observed at the moment of placement. Comparison with accidents
is difficult, because the moment when the accident occurs in relation to a handling
manoeuvre is not regularly reported. Almost all studies and recommendations focus on
the initial phase of handling, namely picking up the load, and the importance of
placement-related problems are ignored.
As mentioned in the introduction, studies focusing on musculoskeletal injuries and
particularly back injuries sometimes classify accidents in two groups, namely, whether
the injury is itself the first and only unanticipated event, or whether it is preceded by
another event, such as slipping (Manning et al., 1984). Based on this distinction, it is
roughly estimated that in one injury in two, and even more, there is no cause: the worker
was doing his normal work (Manning, 1985; Troup et al., 1988). Incident observation
leads to the conclusion that the proportion of accidents that occur without any apparent
reason are possibly overestimated. On the one hand, the notion of “usual” handling is
difficult to define; goods vary all the time, as does the handling context. What is usual is
variability. Furthermore, several of the observed incidents correspond to events that may
be difficult to express in words in an accident declaration. Also, as Nicholson (1985)
emphasizes, the development of data collection systems responds to management goals
and not to prevention goals. The events retained are generally those that are directly

234 M. Lmtie, R. Pelletier /Safety Science 21 (1996) 223-237
related to the injury, such as a loss of balance, a sudden movement, or an impact. Very
few studies reported incidents such as “loss of handgrip” (Troup et al., 19881, control
difficulties and access problems. These frequently observed incidents are possibly rarely
mentioned in published accident analyses not so much because they are never reported,
but rather because this information is not analyzed. Such a classification bias has already
been mentioned by Strandberg (19851, and Manning et al. (1988), for data on loss of
balance.
Up until now, several researchers have speculated on whether the reported circum-
stances depended on the severity of the injury. This question is important from a
practical standpoint because it determines the pertinence of studying minor accidents,
which are more numerous, and by extension, incidents. Currently, there does not seem to
be a clear consensus emerging on this question. For example, some authors retain the
informations on an accident only for those followed by a minimum of three days of
absence. Some have found that the nature of the information collected changes,
depending on whether or not it is a minor accident being reported (Laitinen, 1982;
Niskanen, 1985), while others have noted that the circumstances reported are practically
the same (Laughery and Vaubel, 1993). In the case of incidents, the authors speculated
whether by simply observing them, the main accident risks could be determined, and
whether the conclusions that would be drawn, without knowing the accident picture,
would be correct.
The response to this question is mixed. Environment-related incidents are very
prevalent. They occur, however, most often independently of the specific handling
activity carried out at that moment and they involve most often a material which is
outside the immediate handler’s contact zone. Only 9% of these incidents have a
consequence - for example, a recovery action - which has a potential risk. Their real
importance therefore emerges from the accident analysis. Conversely, it is easy to
understand why activity-related accidents such as “difficulty controlling” or “difficulty
moving” are risky. In fact, 70% of activity-related incidents indicate per se that the
handling involved some difficulties. Accident analysis confirms this interpretation.
However, the importance of balance problems clearly would not have been obvious just
from observing them. Loss of balance situations occurring in a handling activity are
known to be a major source of injuries (Niskanen, 1985; Manning et al., 1984; David,
1985). The study of handling accidents, as well as the interviews with the handlers
(Authier and Lortie, 1993), confirm the importance of the phenomenon. In this study, an
average of 170 handling activities had to be viewed in order to observe some loss of
balance, sometimes just a slight unsteadiness, and more than 5,000 handling activities
had to be viewed to observe a fall. The importance of the problem could not have been
inferred simply by observing these incidents. Consequently, accident analysis remains an
essential step in the diagnosis.
Nonetheless, a few interesting points emerged from observation of these losses of
balance. First, in 75% of the cases (93/121), loss of balance was the first unexpected
event and it was caused by nothing other than the handling activity. In the accident
descriptions, the handler, moreover, omitted mentioning the cause of the loss of balance.
For example, the condition of the floor was only rarely mentioned. However, observa-
tion showed that in 40% of the cases, the handler did not have both feet on the ground.

M. Lortie, R. Pelletier/Safety Science 21 (1996) 223-237 235
At the present time, handling-related loss of balance is mainly analyzed from the
standpoint of slipping: as a result, much attention is paid to the shoe-ground friction
coefficient, which is possibly not the best means to a solution. Furthermore, most of the
observed losses of balance, as well as those described in the accidents, are seen to
correspond poorly to the normally retained triplet of slipping-tripping-falling. These
favoured classification modalities may be preventing these other loss-of-balance circum-
stances from being recognized.
Finally, the sequential nature of the incidents make them difficult to analyze in
accidents. The analysis grid used in this study was developed to make observation easier
in order to achieve a high rate of inter-observer reproducibility in a context were the
activity and the environment are closely related. The differentiation of incidents, namely
whether they are environment- or activity-related, and the introduction of the notion of
consequence, proved to be an effective solution for the observation. This could also
make the analysis of accidents easier, as well as their interpretation. It must be
remember however, that it was a procedural decision; results indicate where incidents
were observed or detected, not their causes.
4.2. Incident observation as a regulation factor
Incidents depend on the handling context (e.g., loading, unloading, arrangement of
goods, type of goods) as well as the loading or unloading strategy developed by each
handler and on his skills. For example, the order in which a handler unloads goods has
an impact on the number of times he will climb and on the spatial context of each
handling manoeuvre. This will then have an impact on the incidents. In loading, some
workers can visualize with remarkable precision the available space, while others
proceed by trial and error. Incident occurrence seems also to depend on the importance
that is given to it by each handler. For example, the placement of goods is much more
systematically precise with some handlers than with others. This may be a question of
skill, but also probably of choice, as it is with instability of equipment. Hence, cart
stability depends, of course, on the supporting surface, but also on the way it is placed
and the way the goods are placed on it, and, therefore, on the way they are handled. All
handlers do not give the same importance to its stability. Some techniques also involve
less exertion, but since they require more skill, they can be associated with more
incidents. Thus, incidents depend simultaneously on the handling context, and on the
strategy and the skills of each worker, but conversely, they also determine the strategies.
Strategies not only have an impact on the “quality” of the handling activity, but also
on the “quantity”. For example, some handlers begin the handling only when they have
a general idea of all of the goods to be loaded. They load the goods only in a
predetermined order. Others load as they proceed, even if it means moving the goods to
a new location later. In unloading, it is the sorting strategy that directly affects the
number of additional handling activities. In both cases, the use of these strategies
implies the existence of important cognitive activities.
Most recommendations are based on experimental studies where everything takes
place without incident. Hence, symmetric grips are almost systematically recommended,
based on biomechanical criteria, even though experimental studies show that these grips

236 M. Lortie, R. Pelletier/ Safety Science 21 (1996) 223-237
are useful for stabilizing the container (Coury and Drury, 1982). However, studies
carried out on handlers show that the latter clearly favour asymmetrical grips (Drury et
al., 1982; Baril-Gingras and Lortie, 1995; Authier et al., 19951, as this offers them major
advantages in controlling and maneuvering the load. This study confirms the pertinence
of these selection criteria. Incident prevention, like recovery, is in fact continuously
involved in the choice of handling technique. Consequently, it seems that accident
prevention cannot be achieved and based on solely biomechanical criteria; incident
prevention or recovery must also be considered.
5. Conclusion
Incidents are an integral part of handling work. Their identification and analysis are
essential in understanding not only the risks associated with handling activities but also
the choices made by the handlers. These choices may involve a handling technique, as
well as the overall strategy. These choices imply the existence of relatively complex
cognitive processes. Unfortunately, the methods recommended and taught take these
aspects very little into consideration.
References
Authier, M and Lortie, M., 1993. Assessment of factors considered to be important in handling tasks by expert
handlers. Int. J. Ind. Erg., 13(11): 331-340.
Authier, M., Lottie, M. and Gagnon, M.. 1995. Manual handling techniques: comparing novices and experts.
Int. J. Ind. Erg. (paper accepted).
Baril-Gingras, G. and Lortie, M., 199.5. The handling of objects other than boxes: univariate analysis of
handling techniques in a large transport company. Ergonomics, 38(5): 905-925.
Coury, B.G. and Drttry, C.G., 1982. Optimum handle positions in a box-holding task. Ergonomics, 25(7):
645-662.
David, G.C., 1985. UK national statistics on handling accidents and lumbar injuries at work. Ergonomics,
28(l): 9-16.
Dnrry, C.G., Law, C.-H. and Pawenski, 8, 1982. A survey of industrial box handling. Human Factors, 24(5):
553-565.
Hoyos, C.G. and Zimelong, B., 1988. Occupational safety an accident prevention. Behavioral strategies and
methods (Elsevier, Amsterdam).
Kjellen, U., 1984. The deviation concept in occupational accident control: 1. Defmition and classification.
Act. Anal. Prev., 16(4): 289-306.
Laughery, K.R. and Vaubel, K.P., 1993. Major and minor injuries at work: Are the circumstances similar or
different? hit. J. Ind. Erg. 12 (4): 273-279.
Laitinen, H., 1982. Reporting non injury accidents: a tool in accident prevention. J. Oct. Act., 42-4):
275-280.
Leplat, J. and Rasmussen, J., 1984. Analysis of human errors in industrial incidents and accidents for
improvement of work safety. Act. Anal. Prev., 16(2): 77-88.
Leplat, J., 1985. Erreur humaine, fiabilite humaine dans le travail (Armand Collin, Paris).
Lottie, M., Lamonde, F., Collinge, C. and Tellier, C., 1996. Analyse des accidents associts au travail de
manutentionnaire darts le secteur du transport. Travail Humain, (paper accepted, expected date of
publication: April).

M. Lortie, R. Pellerier/ Safety Science 21 (1996) 223-237 231
Manning, D.P., Mitchell, R.G. and Blanchfield, P.L., 1984. Body movements and events contributing to
accidental and non-accidental back injuries. Spine, 9f7): 734-739.
Manning, D.P., 1985. Use of an accident model to investigate and record causes of back injuries. Ergonomics,
28(l): 237-243.
Manning, D.P., Ayers, I., Jones, C., Bruce, M. and Cohen, K. 1988. The incidence of underfoot accidents
during 1985 in a working population of 10,000 Meyerside people. J. Oct. Act., 10: 121- 130.
Nicholson, AS., 1985. Accident information from four British industries. Ergonomics, 28(l): 31-43.
Niskanen, T., 1985. Accidents and minor accidents of the musculoskeletal system in heavy (concrete
reinforcement work) and light (painting) construction work. J. Oct. Act., 7(l): 17-32.
Shannon, H.S. and Manning, D.P., 1980, The use of a model to record and store data on industrial accidents
resulting in injury. J. Oct. Act., 3: 57-67.
Strandberg, L., 1985. The effect of conditions underfoot on falling and overexertion accidents. Ergonomics,
28(l): 131-147.
Troup, J.D.G., Davies, J.C. and Manning, D.P., 1988. A model for the investigation of back injuries and
manual handling problems at work. J. Oct. Act., 10: 107-I 19.