Driving Safety Pyramid

 Here's my long standing paradigm for process improvement in general, and safety improvement in particular:

"If you have few incidents or escapes, you can afford to spend time and effort on each to analyze and improve; if you have many, you cannot afford NOT to!"

In my work, and in many US businesses, some version of the safety pyramid has long been used to visualize the safety environment. The history includes work in 1969 by Frank E. Bird, an insurance investigating engineer, who in turn built upon an empirical study by H. W. Heinrich in 1931.  Most safety pyramids look something like this:

 

 

The key point is that serious incidents, like fatalities, do not randomly occur in a vacuum devoid of other factors, but occur in an environment of overlapping risks where a fraction of lower-tier situations align in such a way to generate the more serious tiers.  This is why many of don't like calling car crashes "accidents", and why I do not care for versions of this pyramid which include the term "accidents".  This dynamic is not really "accidental", but a statistical system behaving rather predictably.

Certainly the precise ratios between layers may vary, and if you're informed enough to be have arguments about the "right" ratios, then you're already in pretty good shape. 

Industrial companies have well proven that paying attention to the lower layers of the pyramid to reduce the incidence rate of minor risks will in turn reduce the rate of serious incidents.  You cannot "try harder" or "hope and pray" to reduce deaths alone; you must do the harder work to reduce the overall risk environment and address the frequent but minor issues.

So, how does this apply to car crashes?  Well, the first key point is that many cities do not take action on a situation until deaths or injuries stack up.  This not only needlessly incurs unnecessary deaths and injuries, but it reinforces a passive, reactive view toward street safety.

Industrial companies learned decades ago that in order to improve safety at the top layer, they had to improve visibility and tracing of minor injuries, near misses, and risky situations.  Often this took training, tools, continuous reinforcement and encouragement, and over time creation of a "safety culture" in the workplace.  To make safety improvement took even more work, digging into the information collected and making changes to processes and operating environments to improve safety...often at significant expense.

Essentially all cities and states track motor fatalities.  Many track serious injuries, but few track minor injuries.  Few if any have mechanisms to collect near-misses and hazardous situations, and there is little reason to do so since there are rarely budgets and staff to investigate and address any findings.

This antiquated, reactive, unconcerned approach needs to change.  As with any processes involving complex systems it will take years of pragmatic trial-and-error to create and refine an appropriate pyramid, processes, and tools, but here are a few core concepts:

- Build a model, perhaps a pyramid like this:

- Create a safety policy.  Investigate every death as a critical failure, do root cause analysis, collect lessons-learned, and make changes.  A fatality should be seen as a failure of local gov't and the street design team, with responsibility falling on elected and appointed individuals from the top down.

- Create proactive information-gathering capabilities.  Acquire, market, and reinforce tools, and reward usage, to gain visibility into the lower layers.  EVERY routine cyclist and pedestrian will have a string of personal anecdotes about "almost getting killed" and unreported crashes that resulted in scrapes, bruises, and damage but with no report filed. 

- Have a budget and staff to routinely monitor and mine collected data to identify and address various types of risks.  There can be no sacred cows -- serious re-work of existing system may be warranted!  Plus, it is critical to take action else the work to gain visibility will atrophy.

- When serious underlying issues are identified in one geographical area, look across the whole traffic network and address the same problems throughout.  Do NOT wait for statistical inevitability to manifest, but instead chop a section out of a lower layer and the whole pyramid will shrink! 

- Recognize that culture does not generally change based on reading, talking, and teaching, but through changes in action.  At the outset, some force -- hard pushing -- will be required to generate culture shift.  It will also be hard to gain acceptance for changes on the streets.  But over time, as safety culture expands, it will get easier to make changes, and the reduction in harm will encourage activists.