Autism Prevalence Trends by Birth Year and Diagnostic Year: Indicators of Etiologic and Non-Etiologic Factors - an Age Period Cohort Problem
Abstract/Contents
- Abstract
- The primary objective of this study is to characterize the etiological and non-etiological components of the observed increase in incidence of diagnosis of autism in California from 1980 to 2015. We show that the time trends of autism prevalence by birth year and diagnostic year correspond directly to trends in etiologic (causal) and non-etiologic (non-causal) factors respectively and endeavor to estimate the coefficients of both trends. The primary dataset is incidence of autism diagnosis data from the California Department of Developmental Services (CA-DDS). It provides the numbers of clients newly enrolled for services under an autism classification for each diagnostic year from 1980 through 2015 with separate observations for each birth year and gender. The analysis estimates cumulative incidence to age 10 as a more appropriate measure than prevalence. Knowledge of the birth year and diagnostic year trends is important for elucidating the combined effect of variable etiologic factors, that is, environmental effects broadly defined, which may lead to an understanding of potential prevention and treatment strategies. The birth year trend, controlling for diagnostic year trend, could also be used to predict the future case load of adults with autism needing support, which may inform policy decisions and associated funding requirements for care of these individuals, which already consumes an estimated 1.5% of US GDP. It is straightforward to estimate the sum of the birth year and diagnostic year coefficients, which corresponds to a growth of 12.82% per year from 1980 to 2015, but intractable to estimate the allocation of individual coefficients within that sum. The problem of estimating the birth year and diagnostic year trends falls within the class of age period cohort (APC) problems, because the age factor affects the analysis and the three variables are collinear; there is a lack of identifiability, which prevents reliable estimation of the key variables. We investigated novel methods of analyzing this type of problem and demonstrate a new way to understand the problem. We show that estimating the age factor correctly is both more important and more difficult than indicated in previous APC literature because estimates of the age factor are inherently functions of the coefficients of the period (diagnostic year) and cohort (birth year) effects, which are unknown, and biases in the age factor estimate based on implicit assumptions of these two effects directly affect the resulting estimates.
Description
| Type of resource | text |
|---|---|
| Date created | May 2017 |
Creators/Contributors
| Author | MacInnis, Alexander | |
|---|---|---|
| Primary advisor | Nelson, Lorene | |
| Advisor | Sainani, Kristin | |
| Degree granting institution | Stanford University, Department of Health Research & Policy |
Subjects
| Subject | Stanford Department of Health Research & Policy |
|---|---|
| Subject | autism |
| Subject | prevalence |
| Subject | cumulative incidence |
| Subject | California Department of Developmental Services |
| Subject | birth cohorts |
| Subject | etiologic |
| Subject | non-etiologic |
| Subject | age period cohort |
| Subject | APC |
| Subject | age factor |
| Genre | Thesis |
Bibliographic information
Access conditions
- Use and reproduction
- User agrees that, where applicable, content will not be used to identify or to otherwise infringe the privacy or confidentiality rights of individuals. Content distributed via the Stanford Digital Repository may be subject to additional license and use restrictions applied by the depositor.
- License
- This work is licensed under a Creative Commons Attribution 3.0 Unported license (CC BY).
Collection
Epidemiology & Clinical Research Masters Theses
View other items in this collection in SearchWorksContact information
- Contact
- agm@macinnis.org
Also listed in
Loading usage metrics...