Multi-state models for the analysis of Wheeze in a birth cohort of Western Cape children

Master Thesis

2019

Permanent link to this Item
Authors
Journal Title
Link to Journal
Journal ISSN
Volume Title
Publisher
Publisher
License
Series
Abstract
Introduction Wheezing is common in young children. By the age of six, approximately 50% of children in high-income countries will have experienced at least one episode of wheezing in their life. Furthermore, childhood wheezing may be associated with reduced lung function and increased risk of asthma in later life. Determining the epidemiology of wheeze is complex given that the risk factors vary based on the age of the child and the phenotype of wheeze. Little is known regarding the recurrent nature of childhood wheezing in low- and middle-income contexts. This study aimed to use multi-state models to estimate the rate of transition among various states of wheeze in children from birth to the age of three years. This study also aimed to investigate the association between possible risk factors for childhood wheezing and the estimated transition intensities. Methods The rationale for conducting the study, as well as the objectives of the study, methods and data analysis plan are outlined in the study protocol (Part A). A summary of what is currently known about childhood wheezing is presented as part of the literature review (Part B). The aim of the literature review was to identify known risk factors for childhood wheeze and the methods used to analyse recurrent childhood wheezing, as well as identify the limitations of the current methods used to analyse recurrent childhood wheezing. A manuscript presenting the results of the study is included as Part C. This study was a secondary analysis of data from 1086 children from birth to three years, born to mothers in the Drakenstein area of the Western Cape, South Africa, enrolled at one of two pri- mary care clinics. The data were collected as part of a prospective birth cohort, the Drakenstein Child Health Study. Cox proportional hazards models were used to investigate the association of risk factors with time to first wheezing event and time to recurrent wheezing. Two multistate models investigating the progression of childhood wheezing were constructed. Multiple definitions of childhood wheeze as an outcome were investigated for all constructed models. A simple unidirectional multi-state model and a complex multi-state model with three states (never wheeze, wheeze not associated with lower respiratory tract infection (LRTI), and, lower respiratory tract infection associated wheeze) were constructed. The multi-state model allowed four possible transitions: 1) from “never wheeze” to “wheeze not associated with LRTI” or from 2) “never wheeze” to “LRTI-associated wheeze” or from 3) “wheeze not associated with LRTI” to “LRTI-associated wheeze” and from 4) “LRTI-associated wheeze” to “wheeze not associated with LRTI”. Transition intensities between wheeze states were estimated using discrete time multi-state models. The association of risk factors with transition intensities were estimated using multivariable proportional hazards models. Results Of the 1086 children included in the study, 476 (44%) experienced at least one episode of wheezing, and 227 (21%) experienced more than one episode of wheezing in the first three years of life. A total of 951 episodes of wheezing were recorded in the 36 months of follow-up time. In the multi-state analysis, LRTI-associated wheeze and wheeze not associated with LRTI were equally likely to be the first wheeze event. However, recurrent wheezing events were more likely to follow LRTI-associated wheeze as the first event (0.0020033 vs 8.6683754 × 10−4 ). Male children were at significantly higher risk of experiencing wheeze associated with an LRTI as the first wheezing event and at significantly higher risk of subsequent recurrent wheezing. Children exposed to maternal smoking prenatally had a significantly higher risk of transition to the wheeze state compared to unexposed children. Conclusion Multi-state models provide a novel method for the analysis of wheezing and recurrent wheezing in a cohort of children in South Africa. Multi-state models successfully predicted the progression of children through discrete states of wheeze and produced results consistent with existing literature on childhood wheeze, while accounting for recurrent events and interval-censored data.
Description

Reference:

Collections