This post extends the discussion on Incident Data you can find here – and gives three techniques you can use to analyse your data to produce more helpful information.  Information that will help decision makers in the organisation make choices that reduce the risk profile of your organisation.

To set the scene – As1885.1-1990 Workplace Injury and Disease Reporting Standard – gives a range of fields that should be recorded for Incidents and it is intended to provide a way to compare incident records between organisations.  For that purpose – Incident reporting that follows the standard is quite good – BUT (and this is a big But) – it is not designed for nor intended to be used as a tool for site based decision makers to identify problems / solutions and allocate resources.

Technique 1 – Time Plot of Incidents

This is a surprisingly simple – but quite informative way of viewing your incident data.

Typically incidents are recorded by date and time.  They then would have some look up fields, note the person making the report / investigating the issue – and then provide a short description.  The rate at which incidents / hazard reports are occurring can be observed as the slope of the plot (Incident date on the horizontal axis and the Incident Count (incremented by 1 for each incident) on the vertical axis).  Typically this will look something like:

Incident Trend Plot

This plot shows that there is a steady trend – but with some variations from the line of best fit occurring.  You can “unpack” these variations and try to understand why there was a reduction (or increase) in reports.  This changing trend – which is quick and easy to determine can be a useful metric for your risk management activities on site.  A reduction in reporting can be a sign that team members have “lost heart” with the system – and either due to a lack of feedback or understanding – are no longer reporting near misses.  This can be arrested by a timely intervention by more senior site personnel – showing they care and that there are responses occurring.

Equally an increase or “up-tick” in the rate of reports can be an early warning of an impending more significant incident.

Technique 2 – Bubble Chart Time Plot

Another approach includes a metric that changes the size of the mark at each data point on the Date vs Incident Count plot.  This can be a quick pointer to identify where time plots are diagnostic.  For this data set – there appears to be a correlation between the increase in the rate of incidents and the size of the incident.  After each “spike” in the rate there was a larger fall event.  You can see this in the next image which shows the bubble plot for data filtered for the size of rockfalls at a mine site.  This “up tick” in the frequency of incidents just prior to each major fall can be helpful as a guide for decision makers who need to decide where to deploy personnel and machinery and what activities to focus on.

Bubble Plot Sample

Note – you can apply the same logic to health and safety data – using a nominal figure to match a different outcome.  For example – Near Misses = 1, Minor Injury / First Aid Treatment = 5, Medical Treatment = 20, LTI = 50 and Fatality = 100.  When the incident counts are filtered by loss type (see the next technique section) a view of your history will be a great way of predicting when to apply more resources.

Technique 3 – Incident Data Filtered by Loss Type

Types of loss are often helpful in identifying the types of control that are most required in a business.  Typical incident reporting does not normally do a good job of describing this.  Whilst consequences should also be considered – a distribution of loss types is important to know – and allows a focus on the areas that will matter most.  Consider the next chart – which shows a breakdown of incidents at an operation by major loss type.

Group Incidents

This shows that the single largest loss type is around the Overloading of Human Systems which can be addressed through workplace design, plant selection and provision of suitable PPE / lifting aids.  Applying resources in this area to achieve a 5% change – will lead to bigger results in Incident (and risk) reduction than anything that can be done to improve almost any other area of loss across the business.

Knowing what is happening in the main loss (and control) areas of your business is powerful – it lets decision makers identify more quickly where the “low hanging fruit” are – and can be a better tool for planning campaigns / additional resources than adopting a more predictive / subjective approach based on committee’s or risk assessment team’s perceptions of areas of need.  The change in shape of pie charts month by month is quite instructive and very useful as a comparison both temporally (month to month) and spatially (location to location).

If you’d like to find out more about these approaches and others that Operational Risk Mentoring have successfully applied with organisations – why not contact Peter Standish via phone or email – using the contact details on our home page.

    1 Response to "3 Techniques to Improve Decisions when Looking at Your Incident Data"

    • martin lombe

      this is excellent data i will appreciate sending me some more.

Leave a Reply

Your email address will not be published.