Association Between Alzheimer Dementia Mortality Rate and Altitude in California Counties

Environmental risk factors for Alzheimer dementia are not well understood. Air pollution may have a causative role.¹ In light of findings about lower incidence of psychiatric conditions, such as attention-deficit/hyperactivity disorder, at higher altitudes,² we sought to determine whether rates of dementia were associated with average altitude of residence.

In an effort to minimize differential reporting by location and time, we analyzed the county-specific deaths attributed to Alzheimer dementia reported within a single statewide public health system during 1 year. The California Department of Public Health published deaths attributed to Alzheimer disease in 2005, based on the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision code G30, categorized by county of residence (not of death).³ Sociodemographic information by county was obtained from 2010 US Census data. Mean altitudes of inhabited areas in each county were computed using US Board on Geographical Names data. Given nonlinear effects of altitude on gas physiology, we used the logarithm of altitude in meters as our main predictor, using base 2 to simplify interpretation (each 1-point increase being a doubling in rate).

To account for sociodemographic confounders, we analyzed the association of log₂(altitude) with (1) crude mortality rates, (2) age-adjusted rates, and (3) age-adjusted rates, controlling for each county’s age, sex, race/ethnicity, education, and average income. There may be inconsistencies in reporting of dementia-related mortality, so we also examined (4) a model incorporating total county population and (5) adjusted models including only counties that had more than 15 dementia deaths (which were considered statistically reliable³).

We estimated crude mortality rates as the number of deaths attributed to Alzheimer disease per 100 000 inhabitants aged 65 years and older in each county. Age-adjusted rates were published for each county in the Department of Public Health report.³ We used linear regression to estimate associations. Analyses were performed in Stata (StataCorp). Institutional review board approval was not sought because the epidemiologic data used had already been published in a government report.

The Figure shows log₂(altitude) plotted against age-adjusted dementia mortality rates. A similar trend was seen for the crude mortality rate.

The Table shows the coefficients for log₂(altitude) from the regression models. In age-adjusted models, each doubling in altitude was associated with a roughly 2-point decline in the age-adjusted mortality rate for dementia. The predicted rate at the highest county (1800 m) was 14 points lower than at the lowest county (15 m), or about half. The results were similar in the counties that had more than 15 dementia deaths.

This analysis suggests that altitude of residence may impact the risk for dying of Alzheimer dementia. Ecologic studies of this type must be interpreted cautiously. Unexamined factors may account for differential reporting of cases across counties, including miscategorization of other forms of dementia. For instance, 1 high-altitude county (Alpine) reported no cases in 2005. We were not able to reliably assess many confounders, including comorbidities and air pollution. None of the sociodemographic factors we controlled for mitigated the estimated association, and the same results were found among the larger counties that had more reliable rate estimates.

Altitude of residence might be associated with environmental, lifestyle, or health-related factors, which influence dementia rates. Oxygen levels might have direct long-term effects on brain physiology. Oxidative abnormalities have been long proposed to be central to the pathogenesis of dementia.⁴ One group of researchers previously found that hypoxia prevents neurodegeneration in rats in experimental Alzheimer disease⁵ and hypothesized that adaptation to induced hypoxia may prevent dementia.⁶ To our knowledge, our work is the first to find epidemiological evidence for such effects. Additional work is needed to determine whether this relationship holds in other populations.

Corresponding Author: Stephen Thielke, MD, MS, MA, Geriatric Research, Education, and Clinical Center, Puget Sound VA Medical Center, 1660 S Columbian Way, GRECC S-182, Seattle, WA 98108 (sthielke@u.washington.edu).

Published Online: October 21, 2015. doi:10.1001/jamapsychiatry.2015.1852.

Conflict of Interest Disclosures: None reported.