Clinical and Population Health Research
Center for Infectious Disease and Vaccine Research
First Thesis Advisor
Alan L. Rothman, M.D.
Dengue, Dengue Hemorrhagic Fever, Fever, Child, Thailand, Clinical Laboratory Techniques, Severity of Illness Index, Predictive Value of Tests
Dengue fever (DF) and dengue hemorrhagic fever (DHF) are emerging infectious diseases which are endemic in many regions of the globe, many of which are resource-poor areas. DHF and DF impose a severe economic health burden in tropical and subtropical areas. Dengue virus causes an acute febrile illness that can be a self-limited febrile illness, as seen in most cases of DF, or a life-threatening illness with plasma leakage and shock, as seen in cases of DHF. A systematic review of the literature revealed gaps in the knowledge base of clinical laboratory findings of dengue illness with regards to longitudinal dynamics and classification and predictive modeling of disease severity. The objective of this thesis was to investigate the utility of clinical laboratory variables for classification and prediction of disease outcomes.
The data used in this investigation was derived from a prospective study of Thai children presenting to either of two study hospitals within 72 hours of onset of an acute febrile illness. Systematic data collection, including clinical laboratory parameters, and routine clinical management continued each day until 24 hours after the fever had subsided. A final diagnosis of DHF, DF, or other febrile illness (OFI) was assigned by an expert physician after chart review.
The first research objective of this study was to describe the temporal dynamics of clinical laboratory parameters among subjects with DHF, DF, or OFI. Data were analyzed using lowess curves and population-average models. Quadratic functions of clinical variables over time were established and demonstrated significantly divergent patterns between the various diagnostic groups.
The second research objective was to establish and validate tools for classification of illness severity using easily obtained clinical laboratory measures. Bivariate logistic regression models were established using data from one hospital in an urban area of Thailand as a training data set and validated with a second data set from a hospital in a rural area of Thailand. The validated models maintained a high sensitivity and specificity in distinguishing severe dengue illnesses without using the hallmark indicators of plasma leakage.
The third research objective used classification and regression tree (CART) analysis to established diagnostic decisions trees using data obtained on the day of study enrollment, within the first 3 days of acute illness. Decision trees with high sensitivity were established for severe dengue defined either as: 1) DHF with evidence of shock (dengue shock syndrome, DSS); or 2) DSS or dengue with significant pleural effusion.
This study expands existing knowledge of the potential utility of clinical laboratory variables during different phases of dengue illness. The application of the results of these studies should lead to promising opportunities in the fields of epidemiological research and disease surveillance to reduce the health burden, and improve the clinical management, of dengue illness. Future directions involve application of these algorithms to different study populations and age groups. Additionally, other analytical techniques, such as those involving CART analysis, can be explored with these data.
Potts, JA. Description, Classification, and Prediction of Dengue Illnesses in a Thai Pediatric Cohort: A Dissertation. (2010). University of Massachusetts Medical School. GSBS Dissertations and Theses. Paper 465. DOI: 10.13028/dx4n-qz84. https://escholarship.umassmed.edu/gsbs_diss/465
Rights and Permissions
Copyright is held by the author, with all rights reserved.
Clinical Epidemiology Commons, Health Services Administration Commons, Health Services Research Commons, Infectious Disease Commons, Pathological Conditions, Signs and Symptoms Commons, Virus Diseases Commons