Characteristics of cases and controls are presented in . The median age at diagnosis for cases was 60 years (range, 30–79 years) and the median age at 2 years before interview for controls was 59 years (range, 29–78 years). Diagnosis was made 14.3 months after the onset of symptoms on average. Overall, 27 (25%) of the cases had a bulbar onset and 82 (75%) had a trunk or limb onset. Cases had a median of four different jobs (range, 1–10) before ALS diagnosis. Controls also had a median of four jobs (range, 1–13) up to 2 years before the date of interview ().
Basic characteristics of cases (n = 109) and controls (n = 253).
Construction trades and precision production were both associated with a higher risk of ALS (OR = 2.5; 95% CI, 1.0–5.8 and OR = 2.2; 95% CI, 1.1–4.4, respectively) (). Construction workers excluding supervisors had a 2.9-fold risk of ALS (95% CI, 1.2–7.2), whereas precision metalworking in particular was associated with a 3.5-fold risk of ALS (95% CI, 1.2–10.5). Transportation and material-moving workers had a nonsignificantly higher risk of ALS (OR = 1.9; 95% CI, 0.9–4.3); those using motor vehicles had an even higher risk (OR = 2.2; 95% CI, 0.9–5.6). Service-related occupations, excluding private household and protective service, were associated with a lower risk of ALS (OR = 0.3; 95% CI, 0.1–0.7). The occupational cate gory “farming, forestry, and fishing” was not associated with the risk of ALS (OR = 1.0; 95% CI, 0.2–4.2).
Occupations and the risk of ALS.a
In total, 24 cases (19%) and 49 controls (22%) had served in the military. Military service overall was not associated with the risk of ALS (OR = 1.2; 95% CI, 0.7–2.4). No clear variation of the ORs was noted when military service was subgrouped by branch, duration, or wartime service (data not shown).
shows the association between workplace exposures and ALS. We include for comparison previously reported data showing an association of ALS with lead exposure (Kamel et al. 2002
). Exposure to paint strippers; cutting, cooling, or lubricating oils; antifreeze or coolants; mineral or white spirits; or dry cleaning agents was each associated with a 1.6- to 1.9-fold risk of ALS, although only the association with cutting, cooling, or lubricating oils achieved statistical significance (). Analyses stratified by smoking status showed that ORs were generally larger among non-smokers than smokers, especially for cutting, cooling, or lubricating oils; antifreeze and coolants; and mineral or white spirits (). In an additional model including all three exposures among the nonsmokers, the ORs were 3.9 for cutting, cooling, or lubricating oils; 1.9 for antifreeze and coolants; and 2.9 for mineral or white spirits; none was statistically significant (data not shown). Some interaction with smoking was suggested for antifreeze and coolants (p-
value for interaction = 0.06) as well as cutting, cooling, or lubricating oils (p
-value for interaction = 0.10); none of the other workplace exposures had a significant interaction with smoking (all p-
values for interaction > 0.05).
Self-reported workplace exposures and the risk of ALS, overall and stratified by smoking.a
No exposure except lead (Kamel et al. 2002
) showed a clear-cut dose–response with lifetime days of exposure (). The trend for cutting, cooling, or lubricating oils was also significant (p
= 0.04), but the dose–response curve was not monotonic. Other exposures had a similar nonmonotonic trend, with elevated ORs at medium but not high levels of exposure duration. We found substantial although imprecise elevations in the risk of ALS for 400- to 1,999-day exposure to cutting, cooling, or lubricating oils and for ≥ 2,000-day exposure to dyes or printing inks ().
Self-reported workplace exposures and the risk of ALS by lifetime days of exposure.a
shows the association between ALS and chemicals inferred from the work-place exposures. Aliphatic chlorinated hydrocarbons, ethylene/propylene glycols, glycol ethers, heptane, and hexane were each associated with a 1.5- to 1.7-fold risk of ALS. Analyses stratified by smoking status showed generally larger ORs among nonsmokers than smokers (). The interaction of smoking with exposure was statistically significant for glycol ethers (p = 0.02) and xylene (p = 0.03) and was borderline significant for aliphatic chlorinated hydrocarbons (p = 0.06) and ethylene/propylene glycols (p = 0.06).
Chemical exposures determined by an industrial hygienist and the risk of ALS, overall and stratified by smoking.a
Because workers may have been exposed to several chemicals simultaneously, we examined associations of ALS with each chemical among the nonsmokers while adjusting for the five chemicals associated with ALS in the entire population (aliphatic chlorinated hydrocarbons, ethylene/propylene glycols, glycol ethers, heptane, and hexane; one per each model). Associations between ALS and most chemicals diminished in these analyses except for aliphatic chlorinated hydrocarbons, ethylene/propylene glycols, glycol ethers, and hexane (data not shown).
Exposure to formaldehyde was not associated with the risk of ALS overall (OR = 0.8; 95% CI, 0.5–1.5), and the ORs did not vary greatly by exposure probability (0–1, 1, and 2) or weighted exposure duration (≤ 10,000, 10,001–40,000, and > 40,000 hr) (). An additional analysis was conducted comparing individuals who had a weighted exposure to formaldehyde > 60,000 hr (four cases and four controls) to unexposed individuals, giving an OR of 3.0 (95% CI, 0.7–12.9).
Workplace formaldehyde exposure and the risk of ALS.a
In the interview, the subjects were asked two additional questions: “Did you usually clean your hands with solvents or thinners on any job?” and “Did you ever feel sick or high from an exposure at work?” In total, 78 subjects (26 cases and 52 controls) responded positively to the first question, giving an OR of 1.2 (95% CI, 0.7–2.2), and 57 subjects (16 cases and 41 controls) responded positively to the second question, giving an OR of 0.7 (95% CI, 0.3–1.3).
Because there may still be residual confounding by age in models where it was included as a categorical matching variable, we ran additional models further adjusted for age as a continuous variable. Estimates of the relative risks of ALS were substantially unchanged in these models (data not shown). Finally, we repeated all analyses presented above in three separate sets of sensitivity analyses after excluding a) five subjects whose interviews were characterized as questionable; b) seven cases who had reported a family history of ALS; or c ) 19 cases who had reported a previous trip to islands in the Western Pacific. These analyses did not give substantially changed conclusions (data not shown).