Skip to Main content Skip to Navigation
Master Thesis

Étude des patients hospitalisés lors de l’épidémie de dengue de 2017 en Nouvelle-Calédonie et élaboration d’un modèle prédictif de dengue sévère

Abstract : Introduction: in 2017, New Caledonia experienced a severe dengue epidemic (4486 cases, 15 deaths). The objective of the study was to identify risk factors for severity and to develop a predictive model. Materials and methods: retrospective study of cases of dengue fever (RT-PCR +) hospitalized at the Noumea CHT from January to July 2017. The data were: sex, age, ethnicity, antecedents, treatments, symptoms, biological assessments, presence of warning signs and severity (WHO classification 2009), serotype, dengue and zika IgG serology, medical care. Non-severe cases were compared to severe cases with univariate (Odds ratio, Fisher test) and multivariate (logistic regression) analysis. The adjusted variables for which p <0.2 were used for the construction of a predictive model of risk of severity. Results: among 383 patients included, 299 (78%) showed warning signs, 130 (34%) signs of severity: 22 (5.7%) major hemorrhages, 26 (6.8%) shock syndrome, 54 (14.1%) cytolysis hepatic, 8 (2%) fulminant hepatitis, 38 (9.9%) acute renal failure, 9 (2.3%) myocarditis, 4 (1%) neurological damage. Factors significantly associated with severe forms in univariate analysis were: age group (20-30 years and > 60 years), hypertension, obesity, dyslipidemia, antiplatelet drug use, presence of at least one warning sign, previous infection with a dengue virus. Two predictive models have been developed based on gender. The best models had an AUC of 0.92 and 0.94, Se = 85.7% and 83.3%, Sp = 83.3% and 96.3%, VPP = 70.6% and 90.9%, VPN = 92.6% and 92.9%, Youden score = 0.69 and 0.80, Q coefficient of Yule = 0.93 and 0.98, respectively for women and men models. Conclusion: we report a severe dengue epidemic with a high proportion of liver injury. We constructed a predictive model of severe dengue risk to guide medical care.
Document type :
Master Thesis
Complete list of metadatas

Cited literature [93 references]  Display  Hide  Download

https://dumas.ccsd.cnrs.fr/dumas-01878070
Contributor : Bu Carreire Université de Bordeaux <>
Submitted on : Thursday, September 20, 2018 - 4:02:40 PM
Last modification on : Saturday, September 22, 2018 - 1:13:24 AM
Long-term archiving on: : Friday, December 21, 2018 - 6:37:16 PM

File

Med_generale_2018_Marois.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : dumas-01878070, version 1

Collections

Citation

Ingrid Marois. Étude des patients hospitalisés lors de l’épidémie de dengue de 2017 en Nouvelle-Calédonie et élaboration d’un modèle prédictif de dengue sévère. Médecine humaine et pathologie. 2018. ⟨dumas-01878070⟩

Share

Metrics

Record views

134

Files downloads

2525