Illustration results Clause Samples

Illustration results. An advantage of the growth curve is the ability to model a polynomial in time to better capture the shape of the data. Inclusion of a quadratic effect of time in the model greatly improved model fit (Likelihood ratio test: chi2 136.436, p-value <0.001). Further covariates considered in both models initially are the age of onset (years) and duration of depression (weeks). In Table 4.1, comparing the proposed growth curve model and Weibull model, there are similar patterns in the direction and magnitude of effect. Wald score z-statistics are shown for both models. All the covariate effects are in the same direction, with poorer depression prognosis being associated with more severe baseline clinical presentation. Model results both show a beneficial treatment effect of Escitalopram compared to Nortriptyline; hazard ratio of 0.865 (95% CI: 0.669, 1.119) in the Weibull model. In the GC model, the results are statistically significant and indicate an average decreasing linear trend in the Escitalopram group of -2.268 (95% CI: -2.452, -2.084). This decreasing trend is slightly smaller in the Nortriptyline group, interaction of drug and time 0.157 (95% CI: 0.074, 0.241), indicating an association with more severe MADRS scores in the Nortriptyline group as the study progressed. The depression scores are improving over time, but at a slower rate in the Nortriptyline arm, there is evidence Escitalopram treatment is superior to Nortriptyline in both models. Thus, while effects estimates point to similar effects, in the growth curve model they are substantially more significant. Table 4.1 GENDEP trial dataset. Comparison of estimates from Growth curve, coefficient, and Weibull models, log hazard ratio. * Escitalopram reference category