Factors associated with survival in the National Registry of Veterans with ALS

The clinical course of patients with ALS is highly variable. While the median survival time from symptom onset is 24 years, there are reports of survival ranging from less than a year to more than 40 years. Such variability makes planning difficult for patients and physicians, and complicates clinical trial design. We sought to validate previous predictors of survival and search for new ones using a large group of ALS patients in the National Registry of Veterans with ALS. We were especially interested in how various aspects of military service might affect survival. Subjects were those in the National Registry of Veterans with ALS who had probable or definite ALS (according to El Escorial criteria). A multivariable Cox proportional hazard regression model was used to examine variables for statistical association with ventilator-free survival time (determined from date of first diagnosis). Subjects who had not died or started ventilation by 31 October 2006 were censored. Our group of 1085 US military veterans with ALS was primarily male (98%) and white (94%), with mostly sporadic (95%) and extremity-onset (76%) ALS. Symptom onset occurred at a mean age of 59.3 years (60.6 years for diagnosis). Median survival time from symptom onset was 4.7 years (3.3 years from diagnosis). In our multivariable model, older age at diagnosis (HR 1.41 (95% CI 1.271.55) per 10-year increase), non-extremity site of onset (HR 1.55 (1.241.94)), and past deployment to Vietnam (HR 1.73 (1.362.19)) were all associated with shortened survival. A longer time to diagnosis was associated with better survival (HR 0.77 (0.700.84) per one year increase in diagnosis time). In this unique cohort of veterans with ALS, traditional factors of reduced survival remained important. In addition, past deployment to Vietnam was found to be associated with shortened survival as well. This finding could be due to a common exposure, a shared characteristic, an unmeasured confounder, or an enrollment bias. More research will be needed to understand the reasons behind this new finding.