OBJECTIVES: 1) To measure the prevalence of precipitating events in pediatric chronic pain patients and 2) to compare pain and functional disability outcomes at evaluation and 4-month follow-up by presence and type of precipitating event. METHODS: Precipitating events (e.g., injury) were coded from the medical record for 401 youth (6-19) who presented to a tertiary care chronic pain clinic. Four-month follow-up disability and pain were collected for 187 patients. In addition to frequency of events, we examined differences in pain and disability measures by event type at evaluation and follow-up using multiple statistical analysis strategies. RESULTS: Two-thirds of patients had a precipitating event prior to pain onset. Injury was the most common (55%), followed by chronic disease (23%), infection/illness (12.8%), and surgery (7.5%). Patients whose pain was triggered by injury reported the highest average pain levels, F(3, 340)=2.67, p<.05 and functional disability, F(3, 295)=3.54, p<.05. There were multiple cases of event groups that had significantly different baseline and follow-up psychological measures when compared to the rest of the patient population. Trajectories of pain and disability did not differ between patients with and without a precipitating event. Patients with injuries reported greater improvement in functional disability at follow-up (time x injury) F(1, 183)=4.88, p<.05 whereas patients with chronic disease reported less improvement in disability (time x chronic disease), F(1, 183)=5.49, p<.05. No other interactions were significant for disability or pain. CONCLUSIONS: A majority of patients had experienced some form of precipitating event prior to their pain onset, and the presence of a precipitating event had varied effects on the treatment outcomes of patients at four-month follow-up. Although patients with injuries presented with greater disability and pain, they had significantly more improvement, while chronic disease patients were less likely to improve in terms of functional disability. Type of precipitating event appears to be associated with treatment response and can inform clinical prognoses.

Les données satellitaires représentent aujourd'hui la vaste majorité des observations assimilées dans les modèles de prévision numérique du temps. Leur exploitation reste cependant sous-optimale, seulement 10% du volume total est assimilé en opérationnel. Environ 80% des données infrarouges étant affectées par les nuages, il est primordial de développer l'assimilation des observations satellitaires dans les zones nuageuses. L'exploitation du sondeur hyperspectral infrarouge IASI a déjà permis une amélioration des prévisions météorologiques grâce à sa précision et son contenu en information jamais inégalés. Son utilisation dans les zones nuageuses reste cependant très complexe à cause de la forte non-linéarité des processus nuageux dans l'infrarouge. Cette thèse propose donc une méthode permettant d'exploiter au mieux les observations nuageuses du sondeur IASI. Un modèle de transfert radiatif avancé utilisant les propriétés microphysiques du nuage a été évalué. Cette méthode présente l'avantage majeur d'utiliser les profils de condensats nuageux produits par les modèles de prévision. Grâce à ce nouveau schéma, les profils de contenus en eau nuageuse ont pu être inversés avec succès à partir des observations IASI et d'un schéma d'assimilation variationnelle uni-dimensionnel (1D-Var). L'impact de ces observations en termes d'analyse et d'évolution des variables nuageuses dans le modèle de prévision a aussi été évalué. Cette étude est une première évaluation du choix des variables de contrôle utilisées lors des inversions. Un modèle simplifié uni-colonne du modèle de prévision AROME a permis de faire évoluer les profils analysés par le 1D-Var sur une période de trois heures. Des résultats prometteurs ont montré la bonne conservation de l'incrément d'analyse pendant plus d'une heure et demie de prévision. La formation des systèmes fortement précipitants étant fortement liée aux contenus en eau nuageuse, ces résultats encourageants laissent entrevoir des retombées majeures pour la prévision des évènements de pluie intense et les applications de prévision numérique à très courte échéance. / Nowadays, most data assimilated in numerical weather prediction come from satellite observations. However, the exploitation of satellite data is still sub-optimal with only 10 to 15% of these data assimilated operationally. Keeping in mind that about 80% of infrared data are affected by clouds, it is a priority to develop the assimilation of cloud-affected satellite data. The hyperspectral infrared sounder IASI has already contributed to the improvement of weather forecasts thanks to its far better spectral resolution and information content compared to previous instruments. The use of cloud-affected IASI radiances is still very complicated due to the high non-linearity of clouds in the infrared. This PhD work suggests an innovative way to take advantage of cloud-affected radiances observed by IASI. An advanced radiative transfer model using cloud microphysical properties has been evaluated. This method has the advantage of using cloud water content profiles directly produced by numerical weather prediction models. Thanks to this new scheme, profiles of cloud water contents have been successfully retrieved from IASI cloud-affected radiances with a one dimensional variational assimilation scheme (1D-Var). The impact of these data in terms of analysis and evolution of cloud variables has been evaluated in a numerical weather prediction model. This study is the first step in evaluating the choice that has been made for the control variables used during the retrievals. A simplified one-dimensional version of the AROME model was used to run three-hour forecasts from the 1D-Var analysed profiles. Promising results have shown a good maintenance of the analysis increment during more than one hour and a half of forecast. In regard to these encouraging results, a positive impact on nearcasting applications and forecasts of heavy rainfall events, which are highly coupled to cloud variables, can be expected in the future.