Event Title
The Prevalence of Alzheimer’s Disease : An Application of a Probit Model
Start Date
16-10-2009 2:30 PM
End Date
16-10-2009 4:00 PM
Description
Recent studies have shown an increase in the prevalence of dementia. Alzheimer’s disease (AD) is the most common form of dementia and is defined as the deterioration of memory and other cognitive functions and whose severity is sufficient enough to compromise an individual’s autonomy in their social or professional life. According to the Canadian Alzheimer’s Society, in 2008, Alzheimer’s disease affected one in 20 Canadians over the age of 65 and one in four for those over the age of 85. Therefore, it is understandable that AD is a major concern in the healthcare system (Dartigues et al., 2002).
Studies on risk factors show that age, family history of AD, hypertension, high cholesterol/blood level and low education are associated to the onset of AD. Other factors, such as prolonged use of anti-inflammatory drugs, moderate consumption of red wine or having been to school for eight years or more, can be considered factors that protect an individual against AD. Factors that seem to have an influence in developing AD, include sex or having suffered head trauma but these results do vary from study to study (Lindsay et al., 2002; Dartigues et al., 1991; Arcand-Hébert, 2008).
In the present study, the prevalence of AD in individuals 65 and older is examined by using a probit model referred to in an extensive review of the literature by Maddala (1983) and Greene (2003). The data set used for this study is from the master files of the 2005 Canadian Community Health Survey (CCHS) which contain more than 132 000 observations, and approximately 1900 variables. Within this survey, 30 000 observations of individuals over the age of 64 are available of which 330 participants reported having AD. For this study, all estimations are weighted so that they are representative of the population. We used the Bootstrap method developed by Rao, Wu, et Yue, (1992) and Rust, et Rao, (1996) to ensure the consistency of the estimators.
In the course of this study, the influence of individual characteristics (among others; age, gender, education, ethnic background, place of residence, marital status, weight, other health issues, physical exercise and nourishment) and socioeconomic characteristics (i.e.; activities, revenue, etc.) on the likelihood of being affected or not being affected by AD was examined. We also examined the possibility that certain explicative variables are endogenous and therefore could underestimate the probit model.
Preliminary results show that certain variables are significant at the 5% threshold. The chances of developing AD are associated with having lower education, being 50 years of age or older, being obese and residing in an urban area. Our model also includes variables pertaining to physical exercise, eating habits, tobacco and alcohol consumption and chronic illness. In summary, the probit model will help us determine the influence of these variables on AD. In addition, by using the marginal effects of the probit model, we will be able to better understand the changes in the predicted probability according to AD status.
The Prevalence of Alzheimer’s Disease : An Application of a Probit Model
Recent studies have shown an increase in the prevalence of dementia. Alzheimer’s disease (AD) is the most common form of dementia and is defined as the deterioration of memory and other cognitive functions and whose severity is sufficient enough to compromise an individual’s autonomy in their social or professional life. According to the Canadian Alzheimer’s Society, in 2008, Alzheimer’s disease affected one in 20 Canadians over the age of 65 and one in four for those over the age of 85. Therefore, it is understandable that AD is a major concern in the healthcare system (Dartigues et al., 2002).
Studies on risk factors show that age, family history of AD, hypertension, high cholesterol/blood level and low education are associated to the onset of AD. Other factors, such as prolonged use of anti-inflammatory drugs, moderate consumption of red wine or having been to school for eight years or more, can be considered factors that protect an individual against AD. Factors that seem to have an influence in developing AD, include sex or having suffered head trauma but these results do vary from study to study (Lindsay et al., 2002; Dartigues et al., 1991; Arcand-Hébert, 2008).
In the present study, the prevalence of AD in individuals 65 and older is examined by using a probit model referred to in an extensive review of the literature by Maddala (1983) and Greene (2003). The data set used for this study is from the master files of the 2005 Canadian Community Health Survey (CCHS) which contain more than 132 000 observations, and approximately 1900 variables. Within this survey, 30 000 observations of individuals over the age of 64 are available of which 330 participants reported having AD. For this study, all estimations are weighted so that they are representative of the population. We used the Bootstrap method developed by Rao, Wu, et Yue, (1992) and Rust, et Rao, (1996) to ensure the consistency of the estimators.
In the course of this study, the influence of individual characteristics (among others; age, gender, education, ethnic background, place of residence, marital status, weight, other health issues, physical exercise and nourishment) and socioeconomic characteristics (i.e.; activities, revenue, etc.) on the likelihood of being affected or not being affected by AD was examined. We also examined the possibility that certain explicative variables are endogenous and therefore could underestimate the probit model.
Preliminary results show that certain variables are significant at the 5% threshold. The chances of developing AD are associated with having lower education, being 50 years of age or older, being obese and residing in an urban area. Our model also includes variables pertaining to physical exercise, eating habits, tobacco and alcohol consumption and chronic illness. In summary, the probit model will help us determine the influence of these variables on AD. In addition, by using the marginal effects of the probit model, we will be able to better understand the changes in the predicted probability according to AD status.