Latest / Safe As Podcast / Ep 4: The links between fatal and non-fatal accidents based on >23k incidents

Transcript

Ep 4: The links between fatal and non-fatal accidents based on >23k incidents · Safe As Podcast

• 0:07
  Think about the routine incidents in any high risk
• 0:09
  environment. A tiny gas leak, a minor
• 0:12
  equipment malfunction, A brief loss of control.
• 0:15
  We often treat these as isolated events, easily managed.
• 0:18
  But what if there's an unseen but rather simple thread
• 0:21
  connecting those minor occurrences directly to the
• 0:24
  devastating catastrophes? A connection we've simply failed
• 0:27
  to grasp. Good day everyone.
• 0:29
  I'm Ben Hutchinson. This is Safe Haz, a podcast
• 0:32
  dedicated to the thrifty analysis of safety, risk and
• 0:35
  performance research. Visit safetyinsights.org for
• 0:38
  more research. Today's study is by Linda
• 0:41
  Bellamy titled Exploring the Relationship between Major Fatal
• 0:46
  and non Fatal accidents through Outcomes and Causes, published
• 0:50
  2015 in Safety Science explored the critical link between major
• 0:54
  hazards in both fatal and non fatal accidents.
• 0:58
  To do this, the research analysed over 23,000 serious
• 1:02
  occupational accidents from the Netherlands between 1998 and
• 1:06
  2009. This extensive data was analysed
• 1:09
  using a specialised tool called Storybuilder which use hazard
• 1:12
  specific bow tie diagrams to map out accident causes and effects.
• 1:18
  The study score goal was to directly investigate any
• 1:21
  connections between minor and major accident scenarios.
• 1:24
  So with a bit of background, major action investigations
• 1:27
  often reveal that organisations have types of blind spots and
• 1:31
  overlook in hindsight, warning signs leading to accidents
• 1:35
  incubating in the background, and complex systems have been
• 1:38
  argued to have a type of inherent potential danger.
• 1:42
  However, unlike those massive high profile events that make
• 1:45
  national news, everyday occupational accidents, even
• 1:49
  those with multiple fatalities, rarely get the same public
• 1:52
  attention. The academic discussion on
• 1:55
  accent relationships has also been quite divided.
• 1:59
  Earlier research suggested that smaller accidents are symptoms
• 2:02
  and bigger problems, implying that fixing small ones may help
• 2:05
  to avoid some of the larger ones.
• 2:08
  Conversely, more recent views, especially after major
• 2:11
  industrial disasters like the Texas City refinery explosion,
• 2:15
  argue that occupational injury stats are not reliable
• 2:18
  indicators for process age performance, suggesting a
• 2:21
  complete disconnect. Andrew Hale in 2002 even stated
• 2:25
  that assuming preventing minor accidents automatically prevents
• 2:29
  major ones is based on careless and unsupported reasoning.
• 2:33
  Balamy's study directly tackles these differing viewpoints head
• 2:36
  on. So what did they find?
• 2:39
  Well, the analysis of those 23,000 serious dust accidents
• 2:43
  revealed A crucial insight. Smaller, more frequent accidents
• 2:47
  can provide valuable information about the causes of bigger,
• 2:51
  rarer catastrophic accidents, but only if you're looking
• 2:54
  within the same hazard category. So 36 hazard categories were
• 2:58
  explored in this research included a range of things but
• 3:02
  hazard categories like explosions, falling objects,
• 3:05
  electrical contact, and more. This finding directly
• 3:09
  contradicts the idea that personal and process safety are
• 3:11
  totally unrelated. Instead, Bellamy's research
• 3:15
  concludes that there is a link between occupational and process
• 3:18
  safety and between fatal and non fatal occupational accidents and
• 3:23
  that link is the hazard. Here's my other key results.
• 3:27
  So a commonality in causal or contributing factors across the
• 3:31
  900 plus fatal accidents in this sample there was no barrier
• 3:35
  failure caused unique to fatal accidents that hadn't also
• 3:38
  occurred in non fatal accidents. This strongly suggests that less
• 3:42
  serious accidents of the same hazard type can indicate
• 3:45
  potential for more serious events.
• 3:47
  Lethality inverse frequency. The study used accident ratio
• 3:51
  hills, which will make more sense if you actually read the
• 3:53
  paper, and this shows how hazards differ in their outcome
• 3:56
  severity. Interestingly, the most lethal
• 3:59
  hazards like drowning are not necessarily the biggest killers
• 4:03
  overall. For instance, falls from height
• 4:06
  or contact with falling objects cause far more serious and fatal
• 4:09
  accidents overall, even if other hazards might be statistically
• 4:12
  more lethal per incident. Correlation with hazard types.
• 4:16
  Well, overall general lethality didn't correlate with the accent
• 4:19
  numbers. A significant positive
• 4:22
  correlation was found between non fatal and fatal victims
• 4:25
  within the same specific hazard types.
• 4:28
  This is largely due to the common element of exposure to
• 4:31
  the hazard on barrier failures. Fatal and non fatal accidents
• 4:35
  often share the same underlying barrier failures, but these
• 4:39
  failures don't always occur in the same proportions for
• 4:42
  different severity outcomes. Importantly, there is no one
• 4:46
  fatal barrier failure cause which has not also occurred in
• 4:50
  the non fatal accidents. This suggests a role in breaking
• 4:53
  down incidents at a barrier level.
• 4:56
  So what can we make of these findings?
• 4:58
  Because the underlying contributing and causal factors
• 5:01
  and barrier failures for same hazard accidents can be similar,
• 5:05
  investigating minor frequent incidents for their high
• 5:08
  severity potential and fixing those specific safety barrier
• 5:11
  problems could help prevent bigger, more severe incidents of
• 5:15
  the same time. It's not enough just to look at
• 5:17
  the frequency hazards and their barriers need independent deeper
• 5:21
  scrutiny. Also they provide examples of
• 5:25
  accidents and the types of barrier failures. 1 is the
• 5:28
  American trial scaffold collapsed in 2003.
• 5:32
  Now very quickly this multi fatality accidents underlying
• 5:35
  tributing or underlying causal factors like deficient design
• 5:38
  and anchoring were also found in the less severe non fatal
• 5:42
  scaffold accidents in the database.
• 5:44
  This case study and others illustrate that the causal and
• 5:47
  attributing factors of major incidents are often mirrored in
• 5:50
  seemingly minor ones when the hazard type is the same.
• 5:54
  When I think about this, I think there's a clear finding that
• 5:57
  there can be a connection between fatal and non fatal and
• 6:00
  that connection is the hazard itself.
• 6:03
  Of course, other work and more recent work like that from to
• 6:06
  Hello L and team provides nuance around the magnitude of energy
• 6:09
  and more, suggesting there might be some differences.
• 6:12
  But Simply put, focusing on the hazard scenarios should be one
• 6:16
  key focus. A slip hazard may not be seen as
• 6:19
  fatal, but if it's at height in that scenario, then becomes
• 6:23
  fatal. So what were some limitations?
• 6:27
  It's vital to remember that this study, like any study based on
• 6:30
  accident data, has limitations. The information comes from, in
• 6:34
  an instance, hindsight investigations, which are always
• 6:37
  subject to investigative biases and philtres.
• 6:40
  For example, regulators often focus only on the most severe or
• 6:44
  legal breaches. Also, and this one I think is
• 6:47
  really prevalent, is that what's listed as a causal factor
• 6:51
  depends on the investigator and their own mental models and the
• 6:54
  accident models that they use, and then your reporting
• 6:57
  taxonomies that they follow and a whole range of other pressures
• 7:00
  and influences. Again, what you look for is what
• 7:03
  you find. That's it on Save As.
• 7:06
  I'm Ben Huntington and I hope you found this useful.