This page in Swedish

Laboratory Safety Manual

Investigation of incidents and accidents

When investigating incidents and accidents, working systematically is important to gain a clear understanding of the possible causes of what has happened. Two central questions in the investigation: what happened before the accident, and why did the accident occur? The employer decides which incidents and accidents require investigation in consultation with safety representatives. It is documented in minutes who is to receive feedback in the case, usually the originator of the case, the head of research, the head of division or the safety representative. All employees may also need to be informed, depending on the nature of the case. Once the causes of what occurred are sorted out, preventive measures are taken to stop it from happening again. Those measures not implemented immediately are listed in an action plan, who has the responsibility to ensure they are carried out, and when they should be implemented. The form for investigating accidents and incidents is found under this page’s ”Documents” section. 

Investigation of serious incidents and accidents

Incidents and accidents of a more serious nature require a more extensive investigation. During the investigation, a systematic mapping retroactively is conducted into how the particular incident or accident occurred. The standard course of investigation usually includes the following:

1. Data collection

Methods employed in data collection are photographing the accident scene, sketches of accident scenes, documentation of objects involved in the accident, testimony, interview with experts, measurements, collection of documented instructions, procedures and manuals for objects involved, alarm centre (or equivalent) log lists, a thorough review of the ”accident scene”.

2. Accident analysis

An accident investigation includes everything from the initial planning of how the accident will be investigated, allocation and scheduling of resources, collection and analysis of data and information, recommendations based on the analysis and their implementation, and finally to the evaluation of the effects these recommendations have had. An accident analysis focuses on how an understanding of what has happened can be established based on available data and information. Thus, accident analysis is only one part of an accident investigation. Accident investigation always originates from an accident model. There are a variety of investigative methods. Most methods aim to systematically describe WHAT happened, WHY it happened and what could have PREVENTED it from happening. A systematic approach and methodology strengthen an investigation’s validity and prevent preconceived notions or favourite hypotheses. To ensure that an accident is comprehensively investigated, the applying investigation method must have a system perspective. In an investigation based on a system perspective, the question is asked why an event has occurred in several stages, and not only who did what. A systems perspective’s starting point is the interaction between humans, organisations, and technology, known as “H-O-T”. The basis for a HOT analysis is that human, organisational and technological factors must be given equal attention in an accident investigation. Thus, a H-O-T analysis consists of different sub-analyses:

Incident and cause analysis

An incident analysis aims to reconstruct the course of events and break them down into sub-events. To make the course of events visible and enable links to the other sub-analyses, the reconstruction of the course of events takes place in the form of event diagrams. In the diagram, each sub-event for the course of events is represented by a rectangle with descriptive text. Event diagrams answer the question What happened? Each box should explain the various steps leading to the accident or incident in simple terms. The events are expressed with verbs (walking, pinching, stopping, seeing, falling, etc.) and numbered. Time stamps may also be appropriate. In the next step, at the operational level, one asks Why it happened? Ovals represent causes. To find organisational shortfalls (i.e., the underlying causes), one continues to ask why. Conditions, circumstances, and other influencing factors are often sorted out here. Ovals on several levels can be added: at the departmental level, the corporate level, the community level, etc. For example, the top level is marked with a dash, and above it is found the system level/management level, the controlling level in this context.

Deviation analysis

The deviation analysis is used to identify and document things that, in the various events, deviated from the normal when the incident occurred but did not contribute/cause it. Examples of deviations are: ”Normally there are two doing the assembly, but this time the fitter was alone”; “Normally the assembly is carried out on the dayshift, but this time went into overtime”. The purpose of reporting deviations that have not caused an event is to document that a departure from normal occurred and that the person who performed the analysis evaluated it. In other words, they assessed that it did not contribute to the incident. However, deviations may also be seen as circumstances making it difficult for some to ”do the right thing”.

Barrier analysis

The next step is a barrier analysis. This means discovering barriers that could have stopped the course of events or prevented unwanted consequences from occurring but that have broken or been missing. Ovals may also be used to define why the barriers were missing or broken. Barriers can be of various types, for example, regulatory frameworks, protective devices, etc., and exist during all phases of the course of events.

3. Proposed measures

One of the most important tasks is to develop recommendations or propose measures aimed at preventing the incident from happening again or reducing the risk of a similar incident occurring again. It is common for measures will need to be taken already during the investigation process itself, at both early and late stages. Proposed measures should indicate the problem(s) they intend to solve and a clear link between the root causes and the proposals made. Proposed measures should contain information about What should be done, Who should do it, and When it should be done. The investigation results are documented in a report (according to a template). An investigation reports summary helps the reader to get an overview of what has happened and what is proposed.