Part 2: Improving your safety investigations
March 8, 2010
This is the second in a three-part series.
At some distinct point in the investigation of a particular incident, the investigation team moves beyond the information gathering and assessment and turns toward analyzing that information to determine cause factors. This transition requires a change of mindset. In the information gathering mode, we were open-minded and non-judgmental. As we move to cause analysis, we necessarily become more judgmental. In making this transition to the next step of the investigation, a few more pitfalls come into play.
Incidents generally have more than one cause. If investigations are to deliver the most benefit or value to the organization, you must identify and address all the causes. Your investigation process needs to support the identification and correction of multiple causes and avoid any suggestion that the number should be constrained. Watch for language that asks for the main cause, the top three causes or even something as subtle as having only five lines available on your investigation report form where teams are asked to list identified causes. You will get what you ask for and nothing more.
In cause identification, many organizations fail to utilize a structured cause analysis technique, allowing investigation teams to have complete latitude in what is identified as a cause. In this scenario, human nature and human bias play a large role. Human nature, especially among busy people, will be to think about causes in a narrow and shallow way. Left to our own devices, our thoughts will be colored to a great extent by our biasâ€”how we view the world through our individual lens. The use of a structured cause analysis technique reduces both of these tendencies.
There are a number of commercially available cause analysis tools and other companies have chosen to develop their own internal systems. The intent of all competing systems is the same â€” to force an investigation team to think about causation in a wider and deeper way. The more rigor you use to identify causes creates more targeted and specific corrective actions. An additional benefit to using a structured technique is the common terminology or language that technique will provide, which allows for easier communication about the incident and the lessons you want to share within your organization.
In performing a cause analysis, teams need to carefully consider the behavior of the people involved and why people acted as they did. Such understanding is crucial to prescribing the right corrective actions. The ABC Model (Antecedent- Behavior-Consequence) can be used to create a better understanding of these actions. However, many organizations do not formally review such behavior in investigations, leading to some guesswork when it is time to write corrective actions.
Reports and recommendations
Once the causes are identified, the last remaining task for the investigation team is to prepare a report with their recommendations for corrective action. This stage is where all the value is created â€” the findings are announced and a path forward is set to better manage the conditions and circumstances that allowed the incident to occur. Yet we too often find the report and recommendations are the weakest part of the process. There are two common reasons for this weakness. First, many teams are in a rush to finish. Perhaps they are responding to a rapidly approaching due date, or may simply feel pressure to get back to their normal work, which is generally piling up while they are investigating. Second if there were weaknesses in the information gathering, information assessment or the cause analysis portions, this is where those problems will come home to roost. Extraordinary effort at this stage will not overcome prior deficiencies.
Another potential point of failure is a lack of understanding among report writers as to what traits a recommendation should process. Some organizations use a SMART recommendation approach (SMART representing Specific, Measureable, Attainable, Relevant and Timely). I prefer a somewhat different set of traits.
1) Appropriate scope â€” how widely is this recommendation meant to be applied?
2) Clearly defined endpoint â€” how will the person implementing the recommendation know they are done?
3) Specific and precise wording â€” some words are strong (all, never), use them carefully. Other words are weak (improve, consider) don’t use them as they say nothing.
4) Ability to be closed out â€” can the person responsible actually do this?
5) Aligned with an identified cause â€” what problem is this recommendation meant to address?
Finally, organizations frequently omit two important quality control checks at the end of the investigation. First, have we worked to make existing barriers and prevention strategies more effective before embarking on new efforts? Fixing the existing processes is more effective and economical than creating new ones. Second, the investigation team needs to be able to answer this question without hesitancy â€” if the organization does a workmanlike job in implementing these recommendations, are you confident this event will not recur? If the team cannot respond affirmatively, they still have work to do.
Most investigations don’t deliver full value to the organization, but they fail at these predictable points in the investigation process and in these predictable ways.
Next month: Part three in this three-part series gives you recommendations to improve your incident investigation process.