Le score de propension s'est progressivement imposé comme l’une des méthodes de référence dans l'analyse des données observationnelles afin de prendre en compte le biais potentiel lié à l’existence de facteurs de confusion dans l'estimation de l'effet du traitement sur le critère de jugement. Parmi les recommandations de bonnes pratiques d'utilisation, le processus de sélection des variables à inclure dans le score final utilisé est essentiel, ainsi que l'évaluation de l'équilibre obtenu sur les covariables après appariement ou pondération sur ce score. Dans l'objectif de prioriser l'inclusion et l'équilibre des variables ayant une relation avec le critère de jugement une nouvelle mesure d'équilibre est proposée dans ce travail de thèse. Une première partie de ce travail a eu pour objectif de développer une mesure globale pondérée permettant d'évaluer l'équilibre global des covariables obtenu après appariement et ainsi d'aider à la sélection d’un modèle de propension le plus parcimonieux possible, en éliminant notamment les variables instrumentales. En effet ces variables ne doivent pas être introduites dans le modèle de score de propension au risque de majorer le biais final d'estimation. Lors des étapes d'évaluation de l'équilibre final obtenu, les différentes mesures d'équilibres disponibles ne prennent le plus souvent pas en compte cette information et concluent souvent à l'intérêt d'inclure une telle variable afin de réduire au maximum le déséquilibre entre les groupes. L'évaluation des performances de cette mesure a dans un premier temps fait appel à des simulations de type Monte Carlo. Dans une seconde partie, une mise en application sur des données réelles issues de la médecine d'urgence a permis de préciser la pratique d'utilisation d'une telle mesure / Propensity score (PS) methods have become increasingly used to analyze observational data and take into account confusion bias in final estimate of treatment effects. The goal of the PS is to balance the distribution of potential confounders across treatment groups. The performance of the PS strongly relies on variable selection in PS construction and balance assessment in PS analysis. Specifically, the choice of the variables to be included in the PS model is of paramount importance. In order to priorize inclusion and balance of variables related to the outcome, a new balance measure was proposed in this thesis. First, a new weighted balance measure was studied to help in construction of PS model and to obtain the most parsimonious model, by excluding instrumental variables known to be related with increasing bias in final treatment estimate. Several balances measures are proposed to assess final balance, but none of them help researchers to not include instrumental variables. We propose a new weighted balance measure that takes into account, for each covariate, its strength of association with the outcome. This measure was evaluated using a simulation study to assess whether minimization of the measure coincided with minimally biased estimates. Secondly, we propose to apply this measure to a real data set from an observational cohort study.

Social science researchers are increasingly using multi-group confirmatory factor analysis (MG-CFA) to compare different groups' latent variable means. To ensure that a MG-CFA model is identified, two approaches are commonly used to set the scale of the latent variable. The reference indicator (RI) strategy, which involves constraining one loading per factor to a value of one across groups, assumes that the RI has equal factor loadings across groups. The second approach involves constraining each factor's variance to a value of one across groups and, thus, assumes that the factor variances are equal across groups. Latent mean differences may be tested and described using Gonzalez and Griffin's (2001) likelihood ratio test (LRT[subscript k]) and Hancock's (2001) standardized latent mean difference effect size measure ([delta subscript k]), respectively. Applied researchers using the LRT[subscript k] and/or the [delta subscript k] when comparing groups' latent means may not explicitly test the assumptions underlying the two factor scaling methods. To date, no study has examined the impact of violating the assumptions associated with the two scaling methods on latent mean comparisons. The purpose of this study was to assess the performance of the LRT[subscript k] and the [delta subscript k] when violating the assumptions underlying the RI strategy and/or the factor variance scaling method. Type I error and power of the LRT[subscript k] as well as relative parameter bias and parameter bias of the [delta subscript k] were examined when varying loading difference magnitude, factor variance ratio, factor loading pattern and sample size ratio. Rejection rates of model fit indices, including the x² test, RMSEA, CFI, TLI and SRMR, under these varied conditions were also examined. The results indicated that violating the assumptions underlying the RI strategy did not affect the LRT[subscript k] or the [delta subscript k]. However, violating the assumption underlying the factorvariance scaling method influenced Type I error rates of the LRT[subscript k], particularly in unequal sample size conditions. Results also indicated that the four factors manipulated in this study had an impact on correct model rejection rates of the model fit indices. It is hoped that this study provides useful information to researchers concerning the use of the LRT[subscript k] and [delta subscript k] under factor scaling method assumption violations. / text

The aim of this study was to do a structured analysis of the literature on biomarkers for major depressive disorder (MDD) and how these biomarkers may be modulated by physical activity (PA). This with implications for future treatment of mild to moderate MDD with PA. The method was quantitative and followed guidelines for conducting a simplified meta-analysis. The study analyzed 37 randomized controlled trials (RCTs) that covered a total of 911 individuals doing PA. Articles on biomarkers that are previously well established in their relationship with MDD were collected in a structured way, following strict criteria. Results were achieved using statistical methods for calculating the average effect size (ES) and average mean difference (Δ%) for the biomarkers as a result of PA. BDNF showed an effect size of 0.81 ± 1.09 and an average mean difference of +61.7 ± 112.20 %. CRP showed an effect size of 0.35 ± 0.28 and an average mean difference of -18 ± 13.69 %. Cortisol showed an effect size of 0.09 ± 0.75 and an average mean difference of -2.9 ± 17.30 %. Serotonin showed an effect size of 0.39 ± 0.54 and an average mean difference of -11.53 ± 21.10 %. Testosterone showed an effect size of 0.59 ± 1.46 and an average mean difference of +6.50 ± 20.04 %. The conclusion was that PA had a large effect on BDNF and can be used as a diagnostic- and follow-up tool for patients with MDD treated with PA. PA has a small but consistent effect on CRP which therefore can be used in conjunction with other outcome measures to diagnose and follow up patients with MDD treated with PA. PA showed no effect on cortisol and can therefore be considered to be irrelevant as a diagnostic- and follow-up tool for patients with MDD treated with PA. PA had a small but relatively consistent effect on serotonin which therefore can be used in conjunction with other outcome measures to diagnose and follow up patients with MDD treated with PA. PA had a moderate and relatively inconsistent effect on testosterone which therefore can be used in conjunction with several other outcome measures to diagnose and follow up patients with MDD treated with PA. On the basis of these conclusions PA can be a valuable tool for improving some biomarkers for MDD (BDNF, CRP, serotonin & testosterone). Hopefully this study can provide a basis for further research as well as an addition to first line treatment for mild to moderate MDD with PA. / Syftet med studien var att genomföra en strukturerad analys av litteraturen på biomarkörer och depression (MDD) samt hur dessa biomarkörer kan påverkas av fysisk aktivitet (FA). Detta med implikationer för framtida behandling av mild till måttlig MDD med FA. Metoden var av kvantitativ ansats och följde riktlinjer för en förenklad metaanalys. Studien analyserade 37 randomiserade, kontrollerade studier (RCT) som tillsammans täckte in 911 individer som utförde FA. Artiklar på biomarkörer som tidigare kopplats starkt till MDD samlades in strukturerat enligt strikta kriterier. Resultat erhölls genom statistiska metoder för att beräkna den genomsnittliga effektstorleken (ES) och den genomsnittliga medelvärdesdifferensen (Δ%) för biomarkörerna som ett resultat av FA. BDNF uppvisade en effektstorlek på 0.81 ± 1.09 och en genomsnittlig medelvärdesdifferens på +61.7 ± 112.20 %. CRP uppvisade en effektstorlek på 0.35 ± 0.28 och en genomsnittlig medelvärdesdifferens på -18 ± 13.69 %. Kortisol uppvisade en effektstorlek på 0.09 ± 0.75 och en genomsnittlig medelvärdesdifferens på -2.9 ± 17.30 %. Serotonin uppvisade en effektstorlek på 0.39 ± 0.54 och en genomsnittlig medelvärdesdifferens på 11.53 ± 21.10 %. Testosteron uppvisade en effektstorlek på 0.59 ± 1.46 och en genomsnittlig medelvärdesdifferens på +6.50 ± 20.04 %. Konklusionen var att FA hade en stor effekt på BDNF, vilken därför kan användas för diagnostik och uppföljning av individer med MDD under behandling med FA. FA hade en liten men konsekvent effekt på CRP vilken därför kan användas tillsammans med andra mått för att diagnosticera och följa upp individer med MDD under behandling med FA. FA visade ingen effekt på kortisol som därför kan anses som irrelevant för diagnostik och uppföljning av individer med MDD under behandling med FA. FA hade en liten men relativt konsekvent effekt på serotonin vilken därför kan användas tillsammans med andra mått för att diagnosticera och följa upp individer med MDD under behandling med FA. FA hade en medelstor och relativt inkonsekvent effekt på testosteron och kan därför tillsammans med flera andra mått användas för att diagnosticera och följa upp individer med MDD under behandling med FA. Med bakgrund av dessa slutsatser kan FA vara ett värdefullt verktyg för att förbättra vissa biomarkörer för MDD (BDNF, CRP, serotonin & testosteron). Förhoppningsvis kan denna studie verka som en grund för vidare forskning samt utgöra en ett tillägg till standardiserad behandling för mild till måttlig MDD med FA.