That UC Davis Work Comp Study: PR v. Reality

Thursday, September 29, 2011 | 0

The good folks at the University of California, Davis were kind enough to send me the entire study they are publishing re "...Predictors of Workers Compensation Costs". This is the one that generated headlines claiming financial returns are the best single predictor of work comp premiums.

That's not exactly what the study says. Sort of, but not exactly.

There's a rather large disconnect between the report itself and the press release issued by UC Davis about the report. Moreover the press release itself is misleading, poorly written, and stuffed with quotes that reflect a lack of basic understanding about workers comp. Here's one: "Increasing premiums had nothing to do with the number of injured workers, who often are incorrectly blamed for increasing premiums for employers."

By whom? When and where? This kind of misguided PR flackery is sloppy at best, if not outright harmful. It can, and will, be used to add credibility to and strengthen the position of those with their own unique agendas.

Reading the press release and the report, you'd be hard pressed to know they were about the same underlying research report. The firm, declarative statements in the press release were NOT supported by the much more heavily qualified and less direct statements in the report.

For example, the report said this:

<i>We had two major conclusions. First, the year 1992 marked a sharp contrast in trends and correlations between unemployment and incidence rates for occupational injuries and illnesses. Second, for the entire time period (1973-2007), insurance carriers' premiums were strongly associated with returns on investments.</i>

The press release read thusly:

<i>Skyrocketing workers' compensation claims payments are often blamed for rising premiums, but a UC Davis study has found that the number of claims has dropped during the past two decades... the study shows that higher premiums are instead associated with decreases in the Dow Jones Industrial Average and interest rates on U.S. Treasury bonds. "Insurance companies appear to have been setting premiums according to their returns on the stock and bond markets, not according to the number of claims they have" said J. Paul Leigh, UC Davis professor of public health sciences and senior author of the study.</i>

Note the first sentence especially the phrases "workers' compensation claims payments" and "number of claims". There's at best a marginal connection between the two. As the National Council on Compensation Insurance, the Workers' Compensation Research Institute, the California Workers' Compensation Institute and pretty much every WC actuary has shown uncountable times, the COST of claims is separate and distinct from the NUMBER of claims.

Another problem with the press release the report was about premiums, not costs. There's a BIG difference between the two and conflating them was a serious error.

Leaving aside the big problems with the press release, there's problems with the report too.

As a variable, the researchers selected regular ol' medical inflation as reported by the U.S. Department of Labor. As all of us in the work comp world, trend rates in work comp are NOT the same as trend rates in the rest of the medical payer world. Moreover, we look at medical inflation in two ways on a calendar year and an accident year basis. The researchers said there wasn't much of a correlation between premiums and medical inflation well, given that there's a tail in work comp long enough to circle the block, annual trend just isn't viewed as nor should it be that significant. Which leads one to ask; so, why pick the medical inflation rate as a variable in a study specifically about work comp premium rates?

As I noted yesterday, the report uses U.S. Occupational Safety and Health Administration reportable incidents instead of actual workers comp claims, then conflates the two - repeatedly. That's just outrageously sloppy work. The authors do note that NCCI reports these data, but complains that they don't cover the entire country. So, instead of using ACTUAL REAL workers comp claims, they use another dataset as a proxy and conflate the two without any noticeable effort to correlate the two, identify differences, or account for them statistically. Why didn't the researchers just focus the study on the NCCI states, where there were actual data? If they wanted national information, the researchers could have looked at OSHA data in those states, compared it to claim rates, and come up with some sort of statistical correlation.

That's not all, but I have real work to do.

What does this mean to you?

Don't read too much into press releases, especially if they are as inflammatory as this one was.

And be wary of research conducted by well-meaning folks who seem to think they've discovered something brand new that the rest of us knew existed for decades.

<i>Joe Paduda is co-owner of CompPharma, a consortium of pharmacy benefit managers, and owner of Health Strategy Associates, an employer consulting firm in Connecticut. This column was reprinted with his permission from his Managed Care Matters blog.</i>