Abstract
Background: The evidence-generating process in pharmacovigilance has well-known limitations, and the availability of electronic healthcare data is increasing. Therefore, new methods for signal detection, such as tree-based scan statistics, are emerging.
Objectives: This study aimed to detect potential safety signals following the initiation of selective serotonin reuptake inhibitors (SSRIs) and serotonin-norepinephrine reuptake inhibitors (SNRIs) using tree-based scan statistics. It was investigated whether signals were unknown or listed in the Danish Summary of Product Characteristics (SmPCs). Signals were compared to those found in a previous study using active surveillance with repeated cohorts.
Methods: Using Danish healthcare data from 1996 to 2016, we conducted a data mining study on 15 one-to-one propensity score-matched cohorts of new users of SSRIs and SNRIs. Propensity scores were estimated from a standard set of covariates. Patients were followed for a maximum of 6 months from treatment initiation until an outcome of interest or censoring. Signal detection was performed using the software TreeScan® on incident events on all levels of an ICD-10 diagnosis tree.
Results: In total, 300 unique combinations of exposure, comparator, and medical events were identified. Most of these (82%) were considered unknown. Seven (36.8%) signals suggested for further evaluation in the previous study with repeated cohorts, which were present at the end of the study, were re-detected using tree-based scan statistics. Of these, six were unknown and suggested for further evaluation. For citalopram, these were hepatic failure, ischemic stroke, and renal failure. For escitalopram, they were cardiomyopathy, hemorrhagic stroke (detected as intracerebral hemorrhage), and renal failure.
Conclusions: Tree-based scan statistics were able to detect a large number of signals of adverse events to SSRIs and SNRIs, of which the majority were unknown. Several signals identified in a previous study were confirmed using the tree-based scan statistics.
Objectives: This study aimed to detect potential safety signals following the initiation of selective serotonin reuptake inhibitors (SSRIs) and serotonin-norepinephrine reuptake inhibitors (SNRIs) using tree-based scan statistics. It was investigated whether signals were unknown or listed in the Danish Summary of Product Characteristics (SmPCs). Signals were compared to those found in a previous study using active surveillance with repeated cohorts.
Methods: Using Danish healthcare data from 1996 to 2016, we conducted a data mining study on 15 one-to-one propensity score-matched cohorts of new users of SSRIs and SNRIs. Propensity scores were estimated from a standard set of covariates. Patients were followed for a maximum of 6 months from treatment initiation until an outcome of interest or censoring. Signal detection was performed using the software TreeScan® on incident events on all levels of an ICD-10 diagnosis tree.
Results: In total, 300 unique combinations of exposure, comparator, and medical events were identified. Most of these (82%) were considered unknown. Seven (36.8%) signals suggested for further evaluation in the previous study with repeated cohorts, which were present at the end of the study, were re-detected using tree-based scan statistics. Of these, six were unknown and suggested for further evaluation. For citalopram, these were hepatic failure, ischemic stroke, and renal failure. For escitalopram, they were cardiomyopathy, hemorrhagic stroke (detected as intracerebral hemorrhage), and renal failure.
Conclusions: Tree-based scan statistics were able to detect a large number of signals of adverse events to SSRIs and SNRIs, of which the majority were unknown. Several signals identified in a previous study were confirmed using the tree-based scan statistics.
| Original language | English |
|---|---|
| Publication date | 2022 |
| Number of pages | 2 |
| Publication status | Published - 2022 |