We conducted a case–control study in New England, USA, in 1993–1996 (Kamel et al. 2002
). Sequential ALS cases were recruited from two Boston, Massachusetts, ALS clinics; 71% of eligible cases participated in the study (n
= 110). Diagnosis of ALS by board-certified neurologists specializing in motor neuron disease was based on World Federation of Neurology El Escorial criteria, that is, on the presence of progressive disease with both upper and lower motor neuron signs (Brooks 1994
). Cases were required to live in New England, to be mentally competent, and to speak English. About 85% of cases were enrolled in the study within 1 year of diagnosis, and the remainder within 2 years; 46% were enrolled within 1 year of symptom onset, 73% within 2 years, and 86% within 3 years.
Data collection involved an in-person structured interview and measurements of blood and bone lead; study procedures were completed within 1 month of enrollment. The interview collected information on demographics, lifestyle, occupational history, medical history, and occurrence of ALS in first-degree relatives; its main focus was exposure to lead. We based self-reported occupational exposure to lead on the question, “On any of your jobs, were you exposed 10 times or more to lead in any form (fumes, dust, particles)?” We also recorded dates of first ALS diagnosis and first experience of related symptoms in this interview. We extracted information on respiratory function [forced vital capacity (FVC)] from medical records and available only for a subset of cases (n
= 68). We measured bone lead in tibia and patella using K X-ray fluorescence and blood lead using atomic absorption spectrometry (Kamel et al. 2002
). Blood lead measurements were available for 107 cases and bone lead measurements for 104. We collected whole blood as a source of DNA, and genotyped the K59N polymorphism (rs1800435) in the δ-aminolevulinic acid dehydratase (ALAD
) gene using polymerase chain reaction–restriction fragment length polymorphism (Kamel et al. 2003
For the present study, we obtained information on date and cause of death by searching the National Death Index (NDI) Plus (National Center for Health Statistics, Hyattsville, MD) through 31 December 2003. We considered five items in identifying matches to NDI Plus data: social security number, birth date (day, month, and year), first name, last name, and sex. Seventy-eight individuals were complete matches, with all five items identical. Nineteen were partial matches: either four items were identical, or three items were identical and no more than four digits of the social security number were different. Four poor matches did not meet these criteria, and we found no NDI matches for nine individuals; for 3 of these 13 individuals, death dates were found by searching the records of the Social Security Administration and death certificates were retrieved. Thus, date and cause of death were available for 100 of 110 cases (91%). Comparing the 10 individuals without mortality data with the 100 with data, the former were slightly younger (median age, 57 vs. 61 years) and more likely to be male (8 of 10 vs. 59 of 100), but blood and bone lead levels in the two groups did not differ. ALS was recorded on the death certificate as either an underlying or a contributing cause of death, for 93 of the 100 cases (93%) for whom we had information on date and cause of death.
We used Cox proportional hazard analysis to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) using SAS, version 9.1 (SAS Institute Inc., Cary, NC). We conducted two parallel analyses using as the dependent variable interval from diagnosis to death or interval from symptom onset to death, both defined based on interview data. The 10 cases without mortality data were assumed to be still living and were censored at the end of follow-up; censored individuals contributed person-time to the denominator up to the time of censoring but no events to the numerator. Sensitivity analyses either censoring these individuals at the date of their interview or excluding them gave results similar to those presented below. We censored 7 individuals whose death certificates made no mention of ALS (5 of whom were exact matches) at their date of death, again allowing them to contribute to the denominator but not the numerator. Results were similar if we assumed that these seven individuals had in fact died of ALS and counted them in the numerator.
We modeled blood and bone lead levels as continuous variables after transformation using log2
([Pb] + 32), where [Pb] is lead concentration (Kamel et al. 2002
). Because age, sex, and smoking may be related to both lead exposure and ALS, all models included age at interview (as a log-transformed continuous variable), sex, and a dichotomous variable for ever having smoked; considering smoking status (never, former, current) at the time of the interview gave similar results. Additional factors considered as potential confounders included education (high school or less vs. greater than high school); body mass index (BMI; dichotomized at the lowest quartile) and physical activity level at the time of the interview (hours of sitting, lying down, or sleeping per day, continuous); site of disease onset (bulbar vs. limb); the interval between symptom onset and diagnosis (dichotomized at the median); history of ALS in first-degree relatives; and respiratory function (FVC, dichotomized at the median), assessed in models excluding individuals with missing data. None of these variables changed the effect estimates for lead by > 10%, so none was included in final models.
We considered potential effect modification by age, sex, or interval between symptom onset and diagnosis using stratified models. Too few individuals had bulbar onset, family history of ALS, or diagnosis > 1 year before enrollment for us to consider these groups separately, but we ran models restricted to individuals without these characteristics.
The study was approved by the institutional review boards of participating institutions, and the participants provided written informed consent.