Skip to Main content Skip to Navigation
Master Thesis

Étude épidémiologique rétrospective des noyades en Guadeloupe de janvier 2015 à décembre 2017

Abstract : Rationale: According to the 2012 WHO report, drowning is the third leading cause of accidental death in the world with 372,000 deaths a year. In France, it’s the first cause of accidental death under 25 years old and the second in the population aged from 1 to 4. An original study in the Caribbean zone, Guadeloupe and its dependencies, is justified by the combination of island’s population of more than 400 000 inhabitants, in tropical environment with a practice of aquatic activities in fresh water (rivers, swimming pools) or in salt water, recreational type or not, and with a strong tourist frequentation. Objectives: The objective of our study is to explore the epidemiology of drowning as well as the subphenotypes, in order to develop a prediction model for death. Primary endpoint: binary outcome (‘mortality’) defined by death. Design, setting and participants: From January 2015 to December 2017, a retrospective cohort study is conducted at the University Hospital of La Guadeloupe and its dependencies. It includes all patients (children and adults) admitted to this institution and to the “morgue” for drowning. We apply latent class modelling to identify subphenotype susing clinical data. We model data. Then, we test the association of subphenotypes with clinical outcomes to developa prediction model (after multivariate analysis). Measurements and main results: Among the 171 patient’s admissions in our cohort, 68 (40%) have a complicated outcome, defined as extrahospital death (61 patients, 90%), and intrahospital death(7 patients, 10%). Independent latent class models indicate that a three-class (ie, three subphenotypes) model is the best fit for our cohort; phenotype 1: subjects over 50, malaise in the 300-meter strip of the beach, southern and western region (66% of deaths), phenotype 2: middle-aged subjects, salt water trauma, northern beaches (35% of deaths), phenotype 3: pediatric population, pool drownings (14% of deaths).In multivariate analysis, the main predictors of mortality are the subphenotype, a submersion rate more than 10 minutes, a medical intervention and a Szpilman score up to 2. The resulting prediction model demonstrate good predictive performances in terms of discrimination (c-statistic) 0.9680 [0.942-0.994] and calibration. Conclusions: We identify three subphenotypes among drowning, associated with different mortality rates. One (phenotype 1) of them is characterized by a higher mortality rate above 66%, subjects over 50, fainting within the 300m offshore zone. Identification of drowning subphenotypes could be useful to “personalized «public health message and to inform relatives about the short-term prognosis outcome of drowned patients.
Document type :
Master Thesis
Complete list of metadata

Cited literature [24 references]  Display  Hide  Download
Contributor : Pointe À Pitre Bhu <>
Submitted on : Wednesday, March 11, 2020 - 7:48:50 PM
Last modification on : Monday, July 20, 2020 - 6:46:02 PM


Files produced by the author(s)


  • HAL Id : dumas-02505893, version 1



Marie Paris. Étude épidémiologique rétrospective des noyades en Guadeloupe de janvier 2015 à décembre 2017. Médecine humaine et pathologie. 2019. ⟨dumas-02505893⟩



Record views


Files downloads