On Monday, July 7, 2008, we will hold the first meeting of a new advisory
group that we're bringing together to help us improve our system for identifying
fixed site employers whom we want to inspect.
Our existing system was developing during and after the passage of HB
2830, which laid out the requirement that Oregon OSHA needs to focus on
worksites where we are most likely to encounter hazards. Our current effort
is the first comprehensive attempt to revisit that process.
Our current system and the requirements of HB 2830 are often seen as
being linked in some absolute sense. They are not. The current system
relies on identifying specific employers for inspection based on specific
employer information (claims records). Not only does the law not require
such an approach, but discussion of HB 2830 before the legislative committees
clearly allowed for an approach based on industry risk and industry data,
rather than employer-specific data. In fact, the prime legislative sponsor
of the bill repeatedly referred to such an industry-based approach in
discussing the way he expected the bill's requirements to be implemented.
The law requires us to identify "places of employment," not
"employers." The law specifically allows for "random"
inspections. And the legislative history of the bill explicitly endorses
the possibility of an industry-focused approach.
However, as the stakeholder group that developed the rules tried to improve
the system at the time, it came to focus on the data that was available
- and the most readily available data is claims data. The problem with
relying on claims to identify individual employers is the same problem
we had with letters sent to employers with a "high disabling claims
rate" and the same problem we addressed with safety committees in
proposing HB 2222. Simply put, most employers are too small for one, two
or even three years of claims data to accurately reflect risk. Using claims
data to choose between two otherwise similar employers in the same industry
may be only marginally better, if at all, than choosing between those
employers at random.
In the case of safety, the industries would be ranked and placed in four
risk categories based on various sources of data about injuries and fatalities
within the industry. In this way, we would use both state and federal
information (as well as information both from the Bureau of Labor Statistics
survey and the workers compensation databases). By combining these sources
of data into a single comprehensive ranking, we can identify those places
of employment at the greatest risk based on the nature of the work they
do. After grouping the industries, we would select places of employment
at random from within each group (trying to inspect 30 percent of the
highest risk group each year, and a progressively smaller sample of the
other three groups).
Using this approach (or some variation of it) will enable us to use the
data we have readily available without relying upon employer-specific
claims data to choose between small employers. And it will enable us to
better focus our resources on those industries and activities where we
know the risk of injury is greatest - reflecting the goal of both HB 2830
and of good workplace health and safety practice.
Our approach to health enforcement will be similar, but the nature of
the available data creates some special challenges.