Browsing by Subject "frailty model"

The possibility for patients to renew electronic prescriptions online was introduced in Finland in November 2015. This thesis sets out to answer whether the adoption of this new feature contributed to the medication persistence of antidepressant users by reducing the effort involved in renewing prescriptions. Online renewal can be considered a part of a larger programme to utilise technology to improve Finnish health care, whereas a major necessary intermediary step towards the adoption was the implementation of electronic prescribing. Previous research by Böckerman et al. (2023) documenting an increase in benzodiazepine use following eprescribing serves as an example of the access-improving potential of technology. In addition to describing the institutional setting, this thesis includes a literature review of the economics of medication adherence and health information technology. The methodology section provides an overview of the chosen methods, a mixed effects Cox model with a random intercept, i.e. a frailty model, and a linear probability model with a difference-in-differences estimator and patient fixed effects. While the results of the frailty models without time controls estimated that the hazard of renewal was around three times larger for prescriptions renewable online, this result was contradicted by the last frailty model with time dummies, which suggested that the hazard of renewal for prescriptions renewable online was half of that of prescriptions not renewable online. Both the preliminary inspection of cumulative incidence curves as well as the supplementary analysis with the linear probability model corroborate the notion that online renewal is associated with lower renewal rates. A causal interpretation would be that online renewal had a modest negative effect on the probability of renewal, likely explained by misallocation of online renewal requests to doctors unfamiliar with the patients, resulting in the rejection of those requests. However, given the presence of major underlying trends towards longer-running prescriptions, culminating in the legislative extension of prescription validity at the start of 2017 right after the end of the study period of this thesis, further analysis is needed to verify these findings.