Use this URL to cite or link to this record in EThOS:
Title: Prognostic models in dengue
Author: Lam, Phung Khanh
ISNI:       0000 0004 7965 9392
Awarding Body: Open University
Current Institution: Open University
Date of Award: 2015
Availability of Full Text:
Access from EThOS:
Access from Institution:
Reliable prediction models in dengue would facilitate early identification of patients likely to progress to more severe disease, potentially improving patient management. However, most published studies have limitations with respect to their modelling strategy, sample size, and Chosen clinical outcomes, and to date none have exploited longitudinal data. Moreover, only a few studies have examined outcomes in patients presenting with dengue shock syndrome (DSS), the most severe form of the disease. This thesis aims to overcome these limitations by using two large prospective datasets describing a) 1719 children with established DSS and b) 2598 children hospitalized with dengue. First, the population of children with DSS was characterized, and profound DSS, a composite outcome reflecting the need for intensive supportive care, was established as a suitable outcome for prognostic research in this population. Second, risk factors for profound DSS were identified and included in a robust prediction model. Based on this model, a simple score chart for use in clinical practice was derived. Third, risk factors for progression to DSS among children hospitalized with dengue were identified, and a prognostic model for progression to DSS was carefully developed. However, this model displayed only moderate performance and had limited clinical utility. Lastly, differences between acute and chronic diseases, and the implications for dynamic prediction modeling based on longitudinal data, are discussed. A case study of dynamic prediction modeling for development of DSS suggested that (1) the current platelet count can be used to improve baseline models that rely on enrolment values only, and (2) simple conditional dynamic models displayed similar performance to more complex joint models in this situation.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral