In this post you’ll find a discussion on approaches to analyzing your incident data.  These approaches work – and I would suggest give you a much better understanding of the risks you are facing than trying to sum together the risk scores from multiple qualitative team studies of risk in a business.

How to Analyze Your Data

One important point for me to stress and you to realize – don’t limit your data gathering by pushing for too much analysis too early.

It is better to train line personnel to gather more facts and tell the “story” (with an emphasis on non-fiction) of the incident rather than pressing for analysis too soon.  (For more on this see this earlier post).

The key points to take away on incident data analysis:

  • Assessing the quantity and trend of incidents – with the simplest being to have a cumulative count plotted against date / time of occurrence;
  • Categorizing incidents – using credible major threat types, and;
  • Mapping your experience to the broader industry experience in which you work.

Assessing the Quantity and Trend of Incidents

The Bird triangle represents a useful way to determine whether the incidents you’re hearing about in your organization is a real representation.

Bird Triangle

As an example – most medium sized businesses would suffer 1 to 5 lost workday cases a year (even if this has been managed to a medical treatment level).  This means that there is a target of about 600 Incident reports for the business.  You can do the math for your company – but it is a useful metric to include in your management system to confirm that you are not going to be “blind sided” by a more serious incident.

Alert SymbolNote – the increase in reporting must be supported by senior personnel – to thank people for highlighting issues and to make the experience a positive learning moment rather than an initiator for punishment.

Trend spotting is useful – and one of the easiest ways to do this is to use a simple count function in a spreadsheet view of your data base.  A plot of incident count vs date can show if you are seeing an increase, decrease or consistent set of data.  (Mouse over this image for more or click here to view in a separate browser window).

 

Categorizing Incidents

There is value in gaining a clear understanding of the types of incident that are occurring at your site.

One of the “fatal flaws” in current incident analyses is that there is a focus on the Outcome – not the Incident type in the analysis, reporting and response.

Many major companies have taken this approach – and added another selection field to their incident reports.  I’m not recommending that approach – unless it is applied consistently.  A good way to achieve consistency is to nominate a Role or service provider to review the site’s incident reports – and then apply a category to each incident.  Taking this approach can over-come the noise that develops in most databases – which have similar issues categorized quite differently, which makes the data set less valuable for analysis and nearly impossible for decision makers to trust.

What categories should apply?

The best answer is to apply categories that have common controls within the category – but largely different controls between categories.  A set that ORM have been working with and developed over the past few years is shown in the following graphic.  Some of the bases on which the categories were discriminated included:

  • Organizational arrangements – where certain types of controls (and thus incidents) were dominantly covered by a single department;
  • Where statute or standards mandate a structured control response, and;
  • Commonality of controls – where the measures that Prevent an incident or Mitigate the consequences are relatively unique.

An example of categories that I’ve applied for mining operations is shown below.  This approach though is applicable to any industry sector – replacing or removing any of the 16 elements to match the logical data sets to suit.

These sixteen (16) incident groupings are sufficient to cover all of the losses (with a health and safety focus) in the minerals sector.  (If the descriptive text for each incident type didn’t appear when you moused over – click here – for the source image).  Different groupings work in different industry sectors – but the underlying value of categorizing incident types persists.  As they are logically related to common Causes and Controls (Preventative and Mitigating) the mix of incident types can help to guide decision makers on where the best “bang for buck” is likely to be.

Comparison with Industry Data

If I have seen further than others, it is by standing upon the shoulders of giants.
Isaac Newton

Along similar lines to this classic quote from Sir Isaac Newton – the best lessons we can gain are from losses that we haven’t suffered and solutions developed by others.  Examining what is happening across your industry is powerful – and useful in helping to be more realistic about the risk levels that are reacted to by decision makers.

All of the cognitive biases that people bring to bear make the effectiveness of team sessions at accurately and consistently determining risk levels very doubtful at best.

These are the key factors that form the basis of why I suggest using data – both industry and business for determining risk.  It is still value to gather teams in a “Risk Assessment” format – but use them to:

  • Identify specific local (to the business) issues;
  • Sensitize the team members to issues they could be exposed to;
  • Confirm documented controls are likely to be implemented to arrest or mitigate Risk (or Causal) Pathways – and recommend improvement items where they are identified as weak, and;
  • Achieve consultation on risk related matters across the business (which is a statutory requirement in many jurisdictions).

An example of how I’ve used Industry data is for a recent Broad Brush analysis of risks at a mine site.  After back analyzing over 3000 fatality records the following chart indicates the potential for deaths in their type of mining operation.  This was added to the Risk Pathways that flowed from this (and other) incident data set analyses to produce a library of bow ties – but that is for another story.

The data, when visualized in a simple Pie Chart – was a powerful way of focusing attention on the important incident types – and lifting the decision makers away from the issues with much lower historic significance.

This graphic showed the frequency of major loss types across an entire industry - and was very useful for focusing the decision making team on the most important issues first.

Up next – we’ll look at some other steps that you can take to draw more from your Incident Data and wake the sleeper within!


Leave a Reply

Your email address will not be published.