By a factor of three, the official estimate of work related injury/illness incidence reported in the Bureau of Labor Statistics (BLS) Annual Survey was underestimated for the state of Michigan according to a capture-recapture analysis by Rosenman et al. The study indicated work related disorders impacted 1 out of 5 Michigan workers as opposed to 1 out of 15 as per BLS figures.
Information from 5 state databases that record work related injury/illnesses were matched on a person level and company level for the years 1999, 2000, and 2001. Both computer and manual comparison of data (employee name, company name, employee identification number, social security number, age, nature of disorder, date when occurred, etc.) was performed to optimize the count of all reported claims and create new yearly estimates. These new figures were compared to the state of Michigan information published by the BLS Survey for those years as per the following table.
|
Statistic |
BLS Survey Estimate |
Study Capture-recapture Estimate |
Percent Identified by BLS Survey |
|
Total Average Injury/Illnesses per Year |
281,567 |
869,034 |
32.4% |
|
Total Average Injury/illnesses per Year with Greater than 7 Days Time Loss |
30,800 |
95,033 |
32.4% |
|
Total Average Injuries per year with Greater than 7 Days Time Loss |
26,292 |
80,399 |
32.7% |
|
Total Average Illnesses per year with Greater than 7 Days Time Loss |
4,508 |
14,634 |
30.8% |
The authors suggest BLS Information is compromised as a surveillance system due to:
This article originally appeared in The Ergonomics Report™ on 2006-08-16.