Previous studies that found increased risk for Alzheimer's disease and other age-related dementias may have relied on inappropriate statistical tests, according to a new analysis that found no significant risk when deaths without dementia were not censored but rather were treated as a "competing event." In a cohort of 531 older people followed for a mean 11.5 years, the traditional Cox proportional hazards model showed a hazard ratio for incident dementia of 1.64 (95% CI 1.09-2.46) for the 231 former smokers in the group relative to never-smokers, reported Erin Abner, PhD, of the University of Kentucky in Lexington, and colleagues. The Cox model treats deaths without dementia as a "right censored event," the researchers explained in the Journal of Alzheimer's Disease. But that means early deaths from non-dementia causes aren't taken into account. Instead, Abner and colleagues suggested the so-called Fine-Gray model that treats such deaths as a competing event is a better tool for evaluating risk. Under that model, the researchers found a hazard ratio of 1.21 (95% CI 0.81-1.80) for former smokers versus never smokers. All participants were at least age 60 years, community-dwelling, and relatively healthy without evidence of dementia at study entry from 1989 through 2003. They agreed to be followed annually, including cognitive and neurological assessment, and to donate their brains for autopsy at death. With mean follow-up of 11.5 years, 20.9% of the cohort had been diagnosed with dementia and 45.6% had died without dementia. Median smoking history among former smokers in the cohort was 24.5 pack-years. The cohort also included 49 current smokers; hazard ratios for dementia in this group were (95% CI 0.50-2.87) in the Cox model and 0.70 (95% CI 0.30-1.64) in the Fine-Gray analysis. All analyses were adjusted for baseline age, education, sex, diabetes, head injury, hypertension, overweight, APOE4 genotype, family history of dementia, and use of hormone replacement therapy. The history of smoking and dementia research has been complicated. On the one hand, many (but not all) epidemiological studies had suggested that tobacco use was a risk factor. At the same time, however, nicotine is an acetylcholine receptor activator that -- as long as individuals keep using it -- can act as a memory enhancer. For many years now, drug developers have sought to develop nicotine analogues that stimulate those receptors in the brain more selectively as treatments for cognitive impairment, albeit without much success. Abner noted that, in her group's study, current smoking was a clear risk factor for death without dementia (HR 2.38, 95% CI 1.52-3.72). "The big caveat, however, is that this is just one study, and it is not population based," Abner told MedPage Today. "These [longitudinal] studies are quite burdensome for participants -- they have to come back year after year and agree to donate their brains." Abner said people who agree to participate in them may be healthier and better educated than the average older person. Unexpectedly, autopsy data from 302 deceased participants suggested a lower degree of cerebral pathology (vascular and neurological) in ever-smokers versus non-smokers -- perhaps suggesting that these individuals are not representative of the smoker population at large, and/or that tobacco does have some neuroprotective effect. Nevertheless, Abner said the findings make a compelling case for using competing risk analysis in dementia research moving forward. Earlier work by the University of Kentucky group using the modified Cox modeling suggested that earlier studies linking hypertension to an increased risk for Alzheimer's may also be wrong. One conclusion from their study: the lack of association between current or past smoking with dementia "may have implications for the current focus on smoking cessation" as a means of modifying dementia risk. Funding for this research was provided by the University of Kentucky's Sanders-Brown Center on Aging, the National Institute on Aging, the National Institute of Nursing Research, and others. Source
