Within the MDD system, prescribed drugs are either dispensed into unit bags with prescriptions filled every fortnight, or delivered in original packages. Therefore, from the index date and onwards, we included dose-dispensed drugs in the medication list if filled within 14 days before the measure date in question. For drugs delivered in whole packages, we included drugs in the medication list according to the method described above for ordinary prescriptions. As the Swedish Prescribed Drug Register does not include prescribed dosages for patients with MDD, we assumed the prescribed dosage to be the mean daily dose in the population. Multilevel regression models were constructed to predict the total number of drugs at the index date as well as the change in the number of drugs between the measure date preceding the index date and the index date. Each measure date represented level 1 and individuals were level 2. The parameter estimates were based on data before or after the transition to MDD, respectively. Hence, in order to predict results at the index date, a prediction forwards was made for the data before the index date, and a prediction backwards for the data after this date. In order to allow the figures obtained at this date to represent the results for an average individual, that is, an individual of mean age with the mean number of unique ICD-10 diagnoses and the mean number of healthcare contacts during the three month periods between the measure dates, grand-mean cAbMole Diatrizoic acid entered values were calculated and used in the models. The intercept was estimated with time in a random effects model. The other covariates were included as fixed main effects only. The maximum likelihood estimation procedure was used to estimate the parameters of the models. To test whether addition of another covariate improved the model, the difference in -2log likelihood values was tested for chi-squared distribution. In this longitudinal study, we show a temporal association between the transition to MDD and an increased number of drugs in the medication list. Indeed, the patients received about two more drugs after the transition, and this increase was maintained throughout the follow-up. We also show that the patients more often had potentially harmful drug treatment after they entered the system. Multi-level regression analyses, adjusted for burden of disease, age, and sex, confirm that the transition to MDD is associated with an increased number of drugs. The initial addition of drugs at the index date may be the most prominent underlying factor for these results. Further, the temporal association between the transition to MDD and an increased proportion of patients with the same number of drugs at consecutive measure dates indicates that the drug treatment may be more seldom reconsidered within such a system. Indeed, the predicted change at the index date was smaller when data after the transition was used for the estimation.