Longitudinal data analysis of asthma and wheezing in children | Christian Dehlendorff
| | Abstract | This thesis deals with statistical modelling of asthma and wheezing symptoms in childhood. The main purpose of the thesis is to search for and explore patterns in the occurrence of wheezing in order to understand the development of asthma in childhood.
Asthma has grown to be the most common chronic paediatric illness. The Copenhagen Study of Asthma in Childhood (COPSAC) maintains and collects data from a cohort of high-risk children. The objective of COPSAC is primarily to investigate into the causes of increasing asthma prevalence in society and identify methods for reducing the symptoms and discomforts hereby. Symptom diaries from COPSAC have been analysed in this thesis to nd patterns in the occurrence of wheezing.
The thesis deals with a variety of statistical methods with emphasis on longitudinal data analysis. The applied methods include latent class regression, linear and nonlinear mixed e ects models, generalized estimating equations and logistic regression. The aim of the analysis has been to nd sub-groups (or clusters) of children with the same longitudinal development of wheezing symptoms in order to gain understanding of the dynamics involved in the occurrence of wheezing. The analysis has been performed on di erent response scales and time-scales to investigate the consistency of the results. Furthermore, a comparison between sub-groups and the subsequent asthma diagnosis at the age of 5 years has been done.
Besides the analysis of sub-groups, a number of issues have been addressed in the thesis, including: analysis of seasonal variations in the occurrence of wheezing, analysis of the impact of medication use, and analysis of risk factors with respect to occurrence of wheezing and later diagnosis of asthma. Together these analysis provides insight into the development of asthma in childhood.
Results:
The thesis shows that the children in the cohort can be subdivided into three groups according to their symptom patterns. The three groups are, 1: Children with a high level of symptoms and an increasing symptom-rate until the age of 3 years, 2: Children with a medium level of symptoms initially and a decreasing rate and 3: Children with a low initial symptom-rate and a decreasing or constant rate. Analysis shows that the first group corresponds to the asthmatic group and the low and middle group to the non-asthmatic group. The agreement between group and diagnosis is satisfactory. The results coincide well with important, but sparse, results from literature.
The risk-factor analysis shows that the congenital resistance measured at the age of 1 month is related to the risk of being diagnosed as asthmatic at the age of 5 years. Wheezing and asthma is much more frequent in children with a low congenital resistance compared to children with an above average congenital resistance. No significant risk-factors besides age, season and diagnosis were found for the week to week symptoms. The seasonal component shows that one period corresponds to one year and that the risk of symptoms is higher in winter compared to summer. The seasonal component measures symptoms unrelated to asthma, since the seasonal component is seen to be common for all children.
The thesis shows that accurate predictions of the asthma diagnosis can be obtained by finding patterns in the yearly symptom-rates, since asthmatic and non-asthmatic children have di erent symptom characteristics for the relation between symptoms and age. | | Type | Master's thesis [Academic thesis] | | Year | 2007 | | Publisher | Informatics and Mathematical Modelling, Technical University of Denmark, DTU | | Address | Richard Petersens Plads, Building 321, DK-2800 Kgs. Lyngby | | Series | IMM-Thesis-2007-06 | | Note | Supervised by Assoc. Prof. Klaus K. Andersen, IMM, DTU, and Prof. Per M. Bruun Brockhoff, IMM, DTU. | | Electronic version(s) | [pdf] [ps] | | BibTeX data | [bibtex] | | IMM Group(s) | Mathematical Statistics |
|