States With Low Non-Fatal Injury Rates Have high fatality rates and vice versa

ABSTRACT

Background

State-level injury rates or fatality rates are sometimes used in studies of the impact of various safety programs or other state policies. How much does the metric used affect the view of relative occupational risks among U.S. states? This paper uses a measure of severe injuries (fatalities) and of less severe injuries (non-fatal injuries with days away from work, restricted work, or job transfer–DART) to examine that issue.

Methods

We looked at the correlation between the average DART injury rate (from the BLS Survey of Occupational Injuries and Illnesses) and an adjusted average fatality rate (from the BLS Census of Fatal Occupational Injuries) in the construction sector for states for 2003–2005 and for 2006–2008. The RAND Human Subjects Protection Committee determined that this study was exempt from review.

Results

The correlations between the fatal and non-fatal injury rates were between ?0.30 and ?0.70 for all construction and for the subsector of special trade contractors. The negative correlation was much smaller between the rate of fatal falls from heights and the rate of non-fatal falls from heights. Adjusting for differences in the industry composition of the construction sector across states had minor effects on these results.

Conclusion

Although some have suggested that fatal and non-fatal injury rates should not necessarily be positively correlated, no one has suggested that the correlation is negative, which is what we find. We know that reported non-fatal rates are influenced by workers' compensation benefits and other factors. Fatality rates appear to be a more valid measure of risk. Efforts to explain the variations that we find should be undertaken

*********************************************

From the full-text paper:

This explored the relationship between non-fatal and fatal injuries in the US construction industry, comparing two time periods (2003-2005 and 2006-2008).

• Key findings are in the abstract above, but mainly that across US states there was a negative correlation between non-fatal injuries and fatalities. That is, they moved in opposite directions (states with higher fatalities tended to have lower non-fatal injuries and vice versa).
• There was a clear pattern for southern states to report lower non-fatal injuries and higher fatalities, in contrast to more NW states reporting the opposite trend.
• “The finding that the correlation between non-fatal and fatal injury rates in construction is negative is surprising”, with the authors further noting that “Some may suggest that we should not expect a positive correlation because the causes of fatalities and non-fatal injuries often differ, but it is hard to imagine an argument that the relation between the fatality rate and the true nonfatal rate should be negative” (p8).
• The authors speculate that “non-fatal injuries are relatively underreported in states with high fatality rates” and moreover “Apparently, the level of non-fatal reporting is related to factors that lead to greater prevention efforts” (p8).

• Data indicated that a substantial amount of the variation in non-fatal injury rates was associated with the length of the state’s waiting period (the time before benefits can be received).
• On the above, the “liberality of benefits is positively related to the number of reported injuries” and some evidence suggesting that the higher WC premiums generated by more liberal policies may “increase employers’ efforts at prevention [to reduce costs] and thus help to reduce fatalities” (p9).
• For the general trend of states with higher fatalities but lower injuries, it’s noted that these states tend to be “Southern, to have lower WC benefits, to be less unionized, and to pay lower wages” whereas lower fatalities and higher injury states tend to be “Western, pay higher benefits and wages, are more strongly unionized, and carry out more inspections” (p9) – but the authors advise caution in trying to generalise too much here without further multivariate study.
• They carried out analysis in two separate time periods to reduce the influence of chance on fatality rates (which tend to be rare). Given the consistency of the findings between periods they note that while chance always plays a part, “the relationship needs to be taken seriously” (p9).
• They argue that there should be at least *some* congruence between particular non-fatal and fatal mechanisms/injury types and thus they analysed the relationship with fatal vs non-fatal fall from heights. While the relationship was not as strong as for all injury types, a statistically significant negative correlation still existed (data below).

[Note: Just a proviso, but we have a lot of reasons not to trust reported injury data so caveat emptor on interpreting these findings – although the authors did discuss this issue a little in the paper.]

Link in comments.

Authors: Mendeloff, J., & Burns, R. (2013). American journal of industrial medicine, 56(5), 509-519.