Research: Safety audits don't predict injury

An interesting new article out of Ontario has appeared in the Journal of Safety Research entitled “Developing leading indicators from OHS management audit data: Determining the measurement properties of audit data from the field.”

The crux of the article is that OHS safety audits (which are a tool used in Alberta) don’t have much ability to predict future workers’ compensation claims (claims being a proxy for injury).

Safety audits have long been mooted as possible leading indicators of occupational injuries.

An OHS management audit is used to evaluate the state of a workplace's management structure and processes related to OHS. It determines whether an organization is conforming with a particular standard, such as its own policies and procedures, applicable legislation and regulations, or another standard external to the organization (p.2)

The author’s key hypothesis was safety audit compliance should be inversely related to workers’ compensation lost time claims. Basically, higher compliance scores on the audit should make workplaces safer and thus lower the rate of serious, compensable injuries.

Most of the firms in the sample had very high audit scores. This, the authors suggest. May be an indication of auditor bias towards passing firms (in interesting finding in itself).

The most interesting finding was that, overall, there was no statistically significant relationship between audit scores and claims records (although with enough parsing of the data some relationships could be found). The authors moot that the most likely explanations for the lack of a relationship were “audit score is reflective of recent preparation for the audit and not of true implementation, and… annual claim count data have a large amount of measurement error” (p. 9).

This reflects my own suspicion that safety audits are basically paperwork exercises that have little effect on actual operations and that claims data is an inadequate measure of injury. The upshot is that OHS audits do not, at present, show much hope as leading indicators and should not be viewed as meaningful measures of organizational OHS performance.

-- Bob Barnetsona