A case-referent study on neuropsychiatric disorders among workers exposed to solvents.

A pension clerk in the Swedish province of Örebro is pulling disability files in the early 1970s. Not medical charts or hospital records — pension applications. And in the occupation column, a pattern emerges: painters, varnishers, carpetlayers, again and again, applying for disability due to neuropsychiatric disorders. Axelson, Hane, and Hogstedt observed this pattern and asked a direct question: were the chemicals these workers breathed for decades quietly dismantling their minds? Before their study, the evidence was thin and scattered. Clinicians had long suspected that organic solvents could damage the nervous system over time, but what existed were mostly case reports and small clinical series. The symptoms described were frustratingly vague — memory disturbances, mood changes, headaches, forgetfulness, insomnia, and impaired concentration. Investigators used terms like "nervositas" and unspecified encephalopathic symptoms, highlighting how difficult this syndrome was to define. Painters, varnishers, and carpetlayers consistently appeared in these reports — workers with heavy exposure to turpentine and mixtures of aliphatic and aromatic hydrocarbons, absorbed both through inhalation and skin contact. But without a rigorous study design, it was challenging to separate genuine neurotoxic effects from coincidence, confounding, or diagnostic drift. Axelson and colleagues decided to use the pension register itself as their instrument. They ran a case-referent study, which is also known as a case-control study. Cases were workers who had been considered for a disability pension due to neuropsychiatric disorders. Referents were workers from the same register who received pensions for other reasons and had no history of mental disorder. The register for Örebro, a province of about 275,000 people, covered the years 1969 to 1973 and recorded medical diagnoses, social data, and occupational histories. That last detail made it useful: years spent in specific trades, documented in the same way for everyone in the file. The study sample consisted of 151 cases and 248 referents, all restricted to skilled workers aged 35 to 64. The authors excluded schizophrenia, manic-depressive psychosis, and mental disease of obvious somatic origin, such as dementia following a brain injury. Alcoholism was counted as a case; when it appeared alongside other neuropsychiatric diagnoses, it was taken as the primary diagnosis. Referents had to be entirely free of any mental disorder or social history suggesting one. Exposure was simply defined: years worked as a painter, varnisher, or carpetlayer, with 30 years as the dividing line between lower and higher exposure groups. The central finding is a risk ratio of 1.8. Among the 151 cases, 35, which is 23 percent, had a history in painting, varnishing, or carpetlaying. Among the 248 referents, also 35 had that history, but 35 out of 248 is only 14 percent. This means that solvent-exposed workers were about 80 percent more likely to end up on disability pension for neuropsychiatric reasons than workers in comparable trades who were not exposed. The Mantel-Haenszel estimate, a statistical method for pooling results across age groups, confirmed that ratio at 1.8, with a 95 percent confidence interval of 1.2 to 2.7 when alcoholism cases were included. When alcoholism was entirely excluded, the point estimate remained at 1.8, although the confidence interval widened to 1.0 to 3.2. The association held in both cases. However, the dose-response signal is the most persuasive piece of evidence. When the team split exposed workers by duration — up to 30 years versus more than 30 years — the risk climbed with time. Including alcoholism, the crude risk ratios went from 1.0 in the unexposed group, to 1.5 for shorter exposures, to 2.1 for the longest exposures. When alcoholism was excluded, those ratios were 1.0, 1.3, and 2.3. The average exposure duration in the exposed group was about 30 years. That gradient — more years and more risk, independently of alcohol — is precisely the pattern you’d expect if something in these workplaces was causing the damage. A flat association could be attributed to chance or confounding, while a graded one is much harder to dismiss. Now, Axelson and colleagues are candid about the potential shortcomings. The first issue is diagnostic ambiguity. Pension files provide medical diagnoses "with varying degrees of detail." Diffuse conditions like "nervositas" are inherently weak. They argue that there is little reason to believe that diagnostic deficiencies would cluster specifically in the painting-related occupations. So even if the labels are imperfect, any inaccuracy should be roughly symmetric across exposed and unexposed groups. The second issue is selection. Who ends up in a pension register? Individuals who applied, which means anyone who never qualified as a skilled worker might skew the psychiatric case list. The authors addressed this by restricting both cases and referents to skilled, construction-related occupations, maintaining the comparison within a reasonably homogeneous occupational category. They also noted, interestingly, that painters may even be overrepresented among the referents compared to what population statistics would predict — which would cause the observed risk ratio to skew downward, not upward. If anything, the true association might be larger than 1.8. Third, there’s the question of physical demands. Heavy physical work could drive workers toward somatic disability pensions rather than psychiatric ones, potentially inflating the referent group and artificially raising the risk ratio. The authors examined the occupational breakdown and concluded that the physical demands of painters, varnishers, and carpetlayers were comparable to those in other trades in the study. That’s a judgment call, but it’s a considered one. The one aspect the study couldn't adequately address was the presence of other workplace chemicals. Painters in this era encountered not only hydrocarbon solvents but also pigments, plastic paint components, and wood preservatives. The authors mention that lead exposure among Swedish painters was unusual due to regulations, but they cannot fully eliminate the possibility of contributions from other agents. Only age was controlled through stratification; other potential confounders were not formally adjusted. The authors reported that age introduced only weak confounding because the standardized estimates differed little from the crude ones. However, a modern reader will recognize that one-variable stratification leaves a lot unexamined. Despite these considerations, the core finding survived every sensitivity check the team performed. The risk ratio of 1.8 showed up in the full analysis and in the alcohol-excluded analysis. The dose-response gradient was present in both as well. Chi-square tests yielded p-values of about 0.01 and 0.02. Their conclusion was deliberately cautious: these observations make it likely that solvent exposure in painting, varnishing, and carpetlaying has etiological significance, meaning a real causal role, in chronic neuropsychiatric disorders. It is worth pausing to consider what "etiological significance" meant in 1976. Occupational neurotoxicity was not a well-established field. The notion that routine, chronic exposure to workplace chemicals could result in lasting psychiatric disability — not dramatic poisoning or acute illness, but the slow erosion of cognitive and affective function over a career — was not widely accepted. Demonstrating, with a risk ratio of 1.8 and a clear dose-response, that painters were 80 percent more likely than comparable workers to end up neuropsychiatrically disabled carried significant weight in that context. What makes this paper durable is partly the finding and partly the instrument. Axelson and colleagues didn’t design a clinical trial, recruit a cohort, or measure solvent levels in blood or urine. They used a pension clerk’s files — administrative data collected for an entirely different purpose — and extracted from it an epidemiological signal that a conventional clinical study might never have discovered, because the affected workers had already left the workforce, already been disabled, and already disappeared from the active labor pool where most occupational studies recruit. The register captured them. And what it captured was a pattern of chronic neurological harm in workers whose trades had, for generations, been considered unremarkable. The brain, it turns out, keeps a different kind of ledger than the pension fund does — and eventually, both records agreed. This lecture was created by ennepō. Go to https://ennepo.ai to Discover, Create and Follow the latest research in your field. Read when you can. Listen when you want to.