OBJECTIVES: To model the longitudinal latent effect on self-reported changes in sleep disturbances and disability of adding pregabalin (PGB) to the usual care (UC) therapy of outpatients with GAD treated in routine (Real World) clinical practice.   METHODS: A post-hoc analysis was performed using refractory outpatients with GAD enrolled in a 6-month, prospective study carried-out in psychiatric clinics. Anxiety symptoms, self-reported sleep disturbances and disability changes at end-of-trial visit were assessed with the Hamilton-Anxiety (HAM-A), MOS-sleep and WHO-DAS-II scales, respectively. Adding PGB to the UC was compared to UC alone and its direct and indirect effects (expressed in percentages as the total effect explained by the model) were estimated by means of a conditional latent curve model applying structural equation modeling, with treatment as exogenous variable, and anxiety as covariates. RESULTS: A total of 1546 (68% women, mean age; 45.5 years) PGB naïve patients were included in the analysis. Either symptoms of anxiety, sleep disturbances or disability scores were significantly improved at end-of-trial, with higher scores reduction in the cohort adding PGB: -15.92 vs. -14.48 in HAM-A (F=4.9, p=0.027), -29.67 vs. -23.98 in MOS-sleep (F=16.3, p<0.001), and -23.78 vs. ‑19.71 in WHO-DAS-II (F=24.5, p<0.001), respectively. After discounting the effect on anxiety reduction over time, the estimated PGB effect over UC increase on both sleep disturbances and disability scores were still significant (g=-3.99, p<0.001 and g=-3.07, p<0.001, respectively). Mediation analysis showed that 70% and 58% of the direct effect of PGB over, respectively, sleep disturbances and disability remained after discounting the mediated effect of anxiety improvement. CONCLUSIONS: Adding pregabalin to usual care produced better improvements in sleep disturbances and disability related GAD symptoms in outpatients treated in real world conditions. 70% and 58% of the total effect of PGB over usual care was explained by its direct effect not mediated by improvements in anxiety symptoms.