Rationale and background: Agomelatine (Valdoxan, Thymanax) is a melatonergic agonist and 5-HT2C antagonist indicated for major depressive episodes in adults. Hepatotoxic reactions are an identified risk of agomelatine included in the European risk management plan. The goal of this post-authorisation safety study (PASS) is to evaluate the risk of hospitalisation for acute liver injury (ALI) associated with agomelatine as used in current medical practice in comparison with other antidepressant drugs.
Research question and objectives: Specific objectives will be to estimate the incidence rates of hospitalisation with ALI in adult patients initiating treatment with agomelatine and other antidepressants and to estimate the relative risk of hospitalisation for ALI comparing new users of agomelatine and other antidepressants with new users of citalopram.
Study design: This is a large, multinational, retrospective longitudinal cohort and nested case-control study of new users of agomelatine (main exposure of interest) and new users of citalopram (common reference group), fluoxetine, paroxetine, sertraline, escitalopram, mirtazapine, venlafaxine, duloxetine, and amitriptyline.
Population: The study cohort includes adults from the source populations with at least 12 months of continuous enrolment in the data source who have a first-recorded prescription of agomelatine or one of the other study antidepressants during the study period and had not received a prescription for the same study antidepressant within the prior 12 months. Patients with history of liver disease or risk factors for liver disease, chronic biliary or pancreatic disease; malignancy or other life-threating conditions; and women during pregnancy will be excluded from the study cohort.
Variables: The main exposures of interest will be current use of agomelatine and other selected antidepressants. The primary endpoint, common in all the study data sources, is defined as any patient with a hospital diagnosis for ALI identified with specific ICD-9-CM or ICD-10-CM diagnosis codes. The secondary endpoint is defined by specific and non-specific diagnoses and will be evaluated only in selected study databases (Spain and Denmark) in which validation of this less-specific outcome by review of medical records and/or results from liver tests will be implemented.
Data sources: Based on the results of a feasibility evaluation in 2013, the IACS and SIDIAP databases in Spain, the German Pharmacoepidemiological Database (GePaRD) in Germany, the national registries in Denmark, and the national registers in Sweden have been identified as the best candidate data sources in which to implement the study. All the research institutions except Karolinska Institutet in Sweden have confirmed interest in participating in the study. Researchers in Karolinska will decide during 2015 whether or not to participate.
Study size: The study size is driven by the uptake of agomelatine in the populations from which the automated data sources obtain data. We have estimated that approximately 65,000 to 92,000 users of agomelatine might be available for analysis during the study period. With this study size, and depending on the incidence of hospitalisation for ALI in the study populations, the minimum odds ratio to be detected in the nested case-control study with an 80% power ranges from 2.1 to 6.8 for the scenario with the lowest number of users of agomelatine and from 1.9 to 5.6 for the scenario with the highest number of users of agomelatine.
Data analysis: In the cohort study, crude incidence rates of hospitalisation for ALI will be estimated for current use of agomelatine and each study antidepressant. The Kaplan-Meier method will be used to estimate the crude cumulative incidence of ALI at monthly intervals after the first dispensing of agomelatine and each study antidepressant.
In the nested case-control study, cases and controls will be matched on age, sex, and index date of the event. For both the primary and secondary study endpoints, the risk of ALI in current users of agomelatine and current users of the other study antidepressants will be compared with the risk in current users of citalopram, adjusting for confounders using conditional logistic regression. A secondary analysis will be conducted to estimate the effect in current single users of the study antidepressants. Additional analyses will include the assessment of recent and past use of the study antidepressants.