Latest / Safe As Podcast / E69: Precursors of SIFS & ineffective corrective actions

Transcript

E69: Precursors of SIFS & ineffective corrective actions · Safe As Podcast

• 0:07
  To serious injuries and fatalities, sifts result in
• 0:11
  effective and targeted improvements, not necessarily
• 0:15
  according to this research. Good day everyone.
• 0:18
  I'm Ben Hutchinson and this is Safe as a podcast dedicated to
• 0:22
  the thrifty analysis of safety, risk and performance research.
• 0:26
  Visit safetyinsights.org for more research.
• 0:30
  Today's article is a PhD dissertation from Leslie Stockel
• 0:34
  2023 titled What Causes Accidents at Work?
• 0:39
  Workplace, Serious Incident and Fatality Investigations, A study
• 0:43
  of precursors. This research explored
• 0:46
  contributory factors and precursors and the types of
• 0:50
  corrective actions following serious and fatal accidents and
• 0:54
  their investigations. The study was prompted by a
• 0:58
  quoting the article New and disturbing phenomenon in recent
• 1:02
  years where fatalities in the United States have actually
• 1:05
  risen somewhere around 8.9% from 2021 to 2022, while overall
• 1:12
  workplace injuries have plateaued or decreased after
• 1:15
  decades of steady decline. This perplexing dichotomy
• 1:19
  challenges long accepted accident causation theories that
• 1:22
  primarily attributed accidents to worker behaviours.
• 1:26
  Instead, modern causation theories proposes that Sifs are
• 1:30
  more accurately attributed to a set of operational precursors.
• 1:35
  A SIF precursor is defined as a high risk situation in which
• 1:39
  management controls are either absent, ineffective or not
• 1:43
  complied with, and which will result in a severe or fatal
• 1:47
  injury if allowed to continue. For methods, 150 reports were
• 1:51
  analysed from the NIOSH face reports and also a share and
• 1:55
  internal company reports. The study utilised A2 stage
• 1:59
  mixed method approach, evolving A grounded theory analysis to
• 2:03
  discover the SIFT precursors followed by a content analysis
• 2:07
  to determine their frequency. So what did they find?
• 2:12
  I've skipped a lot. I mean this is a whole PhD
• 2:14
  research dissertation, but I'm going to target a few things
• 2:18
  that I've found particularly interesting.
• 2:20
  I really like the findings around the corrective actions
• 2:22
  following incidents here. Leslie found a significant
• 2:26
  finding was the presence of a disconnect between the
• 2:29
  identified causes of the incidents and the established
• 2:33
  recommendation for corrective or preventative action.
• 2:36
  None of the reports included recommended corrective actions
• 2:40
  that fully addressed the precursor causes for the
• 2:43
  incident. In other words, none of the 150
• 2:46
  sample reports had corrective actions that fully addressed the
• 2:49
  precursors. In fact, 23 didn't address the
• 2:51
  precursors at all and the remaining 127 reports only
• 2:56
  partially addressed the precursors.
• 2:58
  For instance, failure to adopt an effective pre start hazardous
• 3:02
  identification was identified in most reports, yet this was
• 3:06
  rarely specified as a corrective action.
• 3:08
  You know the ways to actually address this apparent
• 3:11
  ineffective or absent pre task identification.
• 3:14
  Specifically, failure of hazard identification was identified as
• 3:17
  a SIFT precursor initiator in 85 instances that was listed as a
• 3:22
  corrective action, something to actually try and resolve in
• 3:24
  future. Only 28 times and trained
• 3:29
  employees and new or revised procedures seem to be used
• 3:33
  almost by default, where in some cases the corrective action
• 3:37
  identified had little or to no correlation to the causes of the
• 3:40
  incident. An example, despite supervising
• 3:45
  instructional condemnation being identified as an organisational
• 3:49
  factor 55 times, supervisor training was only listed as a
• 3:53
  corrective action recommendation four times out of all 150
• 3:56
  reports. Although my view is we do have
• 4:00
  to be rather careful around training, I think these findings
• 4:03
  at least partially support other works like from the Brady review
• 4:06
  of fatal QLD mining accidents. And my research into safety
• 4:11
  auditing also found a disconnect between corrective actions and
• 4:14
  the underlying issue that was being discussed.
• 4:17
  So we found examples where there was a hazard that wasn't
• 4:21
  identified on a risk assessment, let's say, but the fix was to
• 4:25
  run a toolbox talk rather than actually addressing the process
• 4:28
  of risk assessment. Some other notable findings in
• 4:31
  this dissertation. There was a higher proportion of
• 4:34
  incidents occurring in maintenance compared to normal
• 4:37
  operations. Sift precursors for routine and
• 4:41
  non routine may have been miscategorized and, quoting the
• 4:44
  dissertation, poorly defined in previous studies.
• 4:48
  Nevertheless, a higher proportion of serious and fatal
• 4:52
  events in routine work was identified compared to non
• 4:55
  routine. The study clarified that routine
• 4:58
  and non routine should focus on the worker's knowledge and
• 5:01
  experience with the task rather than solely on the facility's
• 5:04
  overall operational frequency. When defined from the worker's
• 5:09
  perspective, routine tasks were identified in more incidents, up
• 5:14
  to 84% of the reports then non routine tasks, a finding that
• 5:21
  refutes other research. Interestingly, Hornagle has
• 5:24
  argued similarly that it's primarily the normal variability
• 5:29
  within routine tasks that can lead to the most severe
• 5:33
  outcomes. 3 major categories of SIF precursors were identified,
• 5:39
  the incident trigger, SMS failures, safety management
• 5:41
  system and organisational failures.
• 5:44
  I've skipped a lot here so this is only just a few points.
• 5:48
  For the initiating mechanism or trigger, this refers to the
• 5:51
  moment a hazard transitions from potential to actual harm.
• 5:55
  This category confirmed precursors like process
• 5:57
  instability, unexpected change, abnormal conditions, safety
• 6:01
  control failures and safety procedure breakdowns was
• 6:05
  commonly implicated. Safety management system
• 6:08
  failures. SMS refers to breakdowns in the
• 6:11
  policies, procedures, practises etcetera identified to help
• 6:15
  mitigate risks. The most significant finding in
• 6:18
  this category was the prevalence of inadequate or absent has
• 6:21
  identification, which was found in 79% of the reports.
• 6:26
  Although it was a proviso like anything, this really depends
• 6:30
  heavily on how the investigators constructed these types of
• 6:33
  findings. You know, what were they
• 6:34
  interested in? How do they perceive the
• 6:36
  causality? For organisational failure, this
• 6:39
  refers to a deterioration in operational processes often
• 6:43
  linked to the organisational climate and leadership.
• 6:46
  Another finding was the supervisor instructional
• 6:49
  condemnation, which I mentioned before.
• 6:51
  So this was said to be a significant finding occurring
• 6:54
  when supervisors either directly instructed unsafe work practises
• 6:58
  or knowingly overlooked condoned employees performing tasks sort
• 7:02
  of improperly or in an unsafe manner.
• 7:05
  Common practise was prevalent across all the data sets,
• 7:08
  indicating a general acceptance among employees and supervisors
• 7:13
  of not following established safety protocols.
• 7:16
  Another sort of nuance here though, is this might be a
• 7:19
  fundamental mismatch with our expectations.
• 7:22
  Maybe we as developers and overlords of the system just
• 7:26
  haven't developed a functional and usable mental model of how
• 7:29
  the work occurs in the 1st place.
• 7:32
  Maybe it's less a problem of people circumventing processes
• 7:35
  and rules and we just didn't have a really calibrated
• 7:38
  understanding of how those things are actually functional
• 7:40
  and usable in reality. But I'm on my soapbox again.
• 7:45
  Previous studies may have also conflated worker experience and
• 7:48
  task frequency. I mentioned this briefly before.
• 7:51
  So this data set found a higher proportion of incidents related
• 7:54
  to workers who experienced in the task.
• 7:56
  In all three data sets, incidents involving workers
• 8:00
  experience with a task was more prevalent than those
• 8:03
  inexperienced with the task. And as it was mentioned, this
• 8:07
  may be because of mismatch between how we actually define
• 8:10
  inexperience. We might be confusing the
• 8:13
  company experience with the task and they might have been doing
• 8:15
  it for decades and conflating that with the actual operators
• 8:19
  that individuals understanding and familiarity with the task.
• 8:23
  52% of the incidents had a some sort of safety procedure
• 8:27
  breakdown expressed in three ways.
• 8:30
  One, an appropriate procedure didn't exist.
• 8:33
  Two, there was a breakdown in the procedure where it may be
• 8:37
  incomplete or poorly implemented.
• 8:39
  And three, the procedure wasn't executed as expected.
• 8:44
  So some of those findings, meh. Even more interesting, and this
• 8:49
  is right on the homestretch, Leslie found that the
• 8:52
  investigators spend an inordinate amount of time, that
• 8:56
  is, they spend a considerable amount of the investigation
• 8:59
  report real estate on describing factors in minute details that
• 9:04
  aren't really contributing to the overall purpose and goal of
• 9:08
  the investigation and ultimately trying to improve.
• 9:12
  For instance, on emergency response actions after the
• 9:15
  incident, it's said that there was a hyper focus on medical
• 9:19
  treatment outcomes for the injured persons and perhaps most
• 9:22
  importantly, a disconnect between incident causes and
• 9:25
  corrective actions. In other words, the reports have
• 9:29
  a hyper focus on describing minute details of the injury and
• 9:33
  its location, treatment and measurements of the sight and
• 9:36
  more. They seem to gloss over, and I'm
• 9:39
  reading a bit between the lines here, but seem to gloss over
• 9:42
  really important evaluations of the precursors of the controls
• 9:46
  and the effectiveness of the controls that work and gloss
• 9:50
  over better quality and goal directed corrective actions.
• 9:54
  These focus areas in the investigation suggest, again by
• 9:58
  Leslie's words, a hyper focus on OSHA reporting and recording
• 10:01
  requirements, which is a common fallacy among employers who use
• 10:06
  OSHA reporting requirements as a organisational measure of safety
• 10:09
  performance. Which get in her words, is a
• 10:12
  problematic and arguably unethical practise.
• 10:15
  The report concludes that these findings align with modern
• 10:19
  safety theories, like from Hornagle's Safety too, which
• 10:23
  proposes shifting focus from just preventing failures to
• 10:27
  understanding what enables successful, efficient, safe
• 10:30
  work. Some of these investigations
• 10:32
  weren't doing. They were dedicating a whole lot
• 10:35
  of the investigation report to discussing the very minute
• 10:39
  stuff, yesterday's incident, not really talking about how to make
• 10:43
  the work safer for tomorrow. And finally, as argued in their
• 10:48
  dissertation, the findings highlight that work is imagined
• 10:52
  versus work is done is a really useful concept.
• 10:55
  So the work is imagined represents formal procedures and
• 10:59
  rules and our expectations and work is done is work is actually
• 11:03
  performed. And again, the investigations
• 11:05
  seem to have a much greater focus on all of these work is
• 11:09
  imagined expectations, but comparatively less focused on
• 11:13
  actually how work was structured and designed and how it occurs
• 11:16
  in practise. So limitations is always
• 11:20
  limitations. Of course I won't cover them
• 11:22
  all, but the usual limit are being based on a very select
• 11:26
  subset of the whole performance spectrum of work.
• 11:29
  What I mean is, serious incidents that were reportedly
• 11:32
  investigated is just a very narrow subset of the overarching
• 11:36
  work. It's the one in 10,000, the one
• 11:39
  in 100,000 times when it doesn't go right.
• 11:42
  And just like history is written by the victors, causes are
• 11:46
  constructed by the investigators, and constructions
• 11:50
  of incidents also include a range of socio political factors
• 11:55
  and the existing mental models of the people constructing the
• 11:58
  meaning of those factors. Whether you intend for it or
• 12:01
  not, it's very difficult to to cleanly and reliably separate
• 12:06
  yourself and your own existing beliefs from the construction of
• 12:10
  causes. That's it on Safe As I'm Ben
• 12:14
  Hutchinson. Please help share, rate and
• 12:16
  review, and check out safetyinsights.org for more
• 12:20
  research. Finally, feel free to support
• 12:23
  Safe As by shouting a coffee link in the show notes.