The study involved 4,134 people (average age 59) who worked at the French national gas and electric company, of whom most worked at the company for their entire career. Their lifetime exposure to chlorinated solvents, petroleum solvents, benzene and non-benzene aromatic solvents was estimated, and they were given the Digit Symbol Substitution Test to assess cognitive performance. Cognitive impairment was defined as scoring below the 25th percentile. Most of the participants (88%) were retired.

For analysis, participants were divided into two groups based on whether they had less than a secondary school education or not. This revealed an interesting finding: higher rates of solvent exposure were associated with cognitive impairment, in a dose-dependent relationship — but only in those with less than a high school education. Recency of solvent exposure also predicted worse cognition among the less-educated (suggesting that at least some of the damage was recoverable).

However, among those with secondary education or higher, there was no significant association between solvent exposure (quantity or recency) and cognition.

Over half the participants (58%) had less than a high school education. Of those, 32% had cognitive impairment — twice the rate in those with more education.

The type of solvent also made a difference, with non-benzene aromatic solvents the most dangerous, followed by benzene solvents, and then chlorinated and petroleum solvents (the rates of cognitive impairment among highly-exposed less-educated, was 36%, 24%, and 14%, respectively).

The findings point to the value of cognitive reserve, but I have several caveats. (Unfortunately, this study appears in a journal to which I don’t have access, so it’s possible the first of this at least is answered in the paper.) The first is that those with less education had higher rates of exposure, which raises the question of a threshold effect. Second is that the cognitive assessment is only at one point of time, lacking both a baseline (do we know what sort of average score adults of this age and with this little education would achieve? A quick online search threw up no such appropriate normative data) and a time-comparison that would give a rate of decline. Third, is that the cognitive assessment is very limited, being based on only one test.

In other words, the failure to find an effect among those with at least a high school education may well reflect the lack of sensitivity in the test (designed to assess brain damage). More sensitive tests, and test comparisons over time, may well give a different answer.

On its own, then, this finding is merely another data-point. But accumulating data-points is how we do science! Hopefully, in due course there’ll be a follow-up that will give us more information.