Rationale and background
Benzodiazepines are commonly prescribed for their anxiolytic, hypnotic, and sedative effects. Despite the use of benzodiazepines during pregnancy, there is limited evidence to support their use during this period or to favour their use over alternative treatments that may provide similar symptom relief with differing safety profiles. Understanding the patterns of benzodiazepine use during pregnancy in Europe, together with the rates of pregnancy losses, is essential for evaluating safety and effectiveness. Despite detailed pregnancy information in many data sources, pregnancy episodes in electronic health record (EHR) data are often inconsistently coded across sources.
As part of the upcoming benzodiazepines periodic safety update report single assessment (PSUSA), the Pharmacovigilance Risk Assessment Committee (PRAC) has requested real-world evidence (RWE) on the utilisation of commonly used benzodiazepines during pregnancy. Additionally, the background rates of pregnancy losses will be described to help contextualise the assessment of treatment safety during pregnancy. To date, two data partners within the DARWIN EU® Data Network have pre-processed pregnancy episodes and developed a Pregnancy Extension Table (PET). While the table has been successfully employed in other contexts, this study marks the first application of this table within the DARWIN EU® Data Network.
Research question and objectives
Research Question: This study is meant to inform on the use of benzodiazepine among pregnant individuals and the feasibility of a risk-assessment study that investigates benzodiazepine use during pregnancy.
The specific study objectives are the following:
1. To characterise users of benzodiazepine and alternative treatments (SSRIs, SNRIs, Z-hypnotics, and Melatonin) during pregnancy in terms of demographics, prior medications, history of mental illness and other comorbidities.
2. To characterise treatments with benzodiazepine and alternative treatments during pregnancy in terms of duration, posology, and indication of prescription during pregnancy.
3. To describe the prevalence of benzodiazepine and alternative treatments’ use during pregnancy
4. To describe trajectories of prescriptions fills for benzodiazepine and alternative treatments throughout the year before pregnancy, pregnancy period, and one month following pregnancy end date.
5. To estimate the incidence of pregnancy loss among all pregnancies and in benzodiazepines and alternative treatment users during pregnancy (when numbers allow).
6. To characterise individuals with pregnancy loss in terms of demographics, comorbidities, and prior medications.
Objectives 1, 2, 5, and 6 will also include stratification by age categories and pregnancy periods. When counts allow, objectives 2, 3, 4, and 5 will be stratified by each benzodiazepine active ingredient, grouped by
MODEL DE SOL·LICITUD
2 IMP-126-CT Versió 07
benzodiazepines’ half-life (short and long-acting). When referring to alternative treatments, each alternative treatment will be reported separately (except for objective 4).
Exploratory objectives (to assess their suitability for subsequent analyses)
7.To estimate the incidence rates of potential negative control outcomes (e.g., musculoskeletal injuries, skin conditions, urinary tract infection) in benzodiazepine and alternative treatment users.
Methods
Study design:
Cohort study design.
Study Period:
From 01/01/2010 until 31/12/2023 or the first and last date of data availability in each database.
Population:
Individuals of female sex (at birth) and at least one year of prior database history, with a pregnancy episode during the study period (2010-2023) and a pregnancy start date on or before December 31, 2022. For specific objectives, the following nested cohorts will be defined as follows:
Objectives 1, 2, and 3: Individuals exposed to drugs of interest during pregnancy. For objectives 1 and 2, only the first exposure episode will be considered.
Objective 4: Individuals exposed to drugs of interest at any point during the year preceding pregnancy, pregnancy, or within 1 month following pregnancy end date.
Objective 5: No subset will be required. The main analysis will not impose any exclusion for a prior history of pregnancy, while additional analysis will: 1. exclude anyone with a pregnancy history in the year before the pregnancy start date. 2. exclude anyone with an unknown pregnancy outcome.
Variables
Indications: Indications will be derived from data by the presence of at least one diagnostic code for the following conditions (assessed using different time windows relative to index date): anxiety disorders, sleep disorders, including insomnia, and history of mental illness (a more general group of ” mental illnesses” [including depression, bipolar disorder, schizophrenia and psychotic disorders, excluding anxiety and insomnia/sleep disorders]).
Exposures of interest:
– Benzodiazepines (at the active ingredient level). Prescription of any benzodiazepine, defined as the presence of any RxNorm codes.
Alternative treatments:
– Selective serotonin reuptake inhibitors (SSRI) (at the active ingredient level). Prescription of any SSRI, defined as the presence of any RxNorm codes.
– Selective noradrenaline reuptake inhibitors (at the active ingredient level). Prescription of any SNRI, defined as the presence of any RxNorm codes.
– Z-hypnotics (at the active ingredient level). Prescription of any Z-hypnotic, defined as the presence of any RxNorm codes.
– Melatonin (at the active ingredient level). Prescription of Melatonin, defined as the presence of any RxNorm codes.
Outcomes of interest
This study will describe the characteristics of benzodiazepine and alternative treatment users, treatment patterns (including first treatment era duration and dose), and indications. The prevalence of benzodiazepine and alternative treatment use will be estimated. Prescription trajectories categorised as restarting, switching, restarting with switching to another, or discontinuation will be assessed. The incidence of pregnancy loss (miscarriage and stillbirth combined, and separately) will be estimated, and individuals
who experience these events will be characterised. Additionally, the study will estimate the incidence of potential negative control outcomes, such as musculoskeletal injuries, skin conditions, and urinary tract infections, among benzodiazepine and alternative treatment users.
Relevant covariates
Pregnancy start and end date, gestational week, pregnancy year (pregnancy start date), age groups (=24, 25-29, 30-34, =35), selected conditions and medications, number of healthcare visits, prior pregnancies, and pregnancy period (<20, >20 weeks).
Follow up
The index dates and follow-up will be different for the cohorts of interest and objectives and will consist of:
For the general population cohort, individuals will be followed from the first date of eligibility criteria fulfilment (pregnancy start date), whereas for the benzodiazepine and alternative treatment nested cohorts, follow-up will start at the date of first treatment initiation during pregnancy (index date). Follow-up will end with pregnancy end date (irrespective of the outcome), or censoring (due to loss to follow-up, death), whichever occurs first.
For objective 4, follow-up will begin at the first initiation of the treatment of interest (index date) within the period contained within the year prior to pregnancy start, pregnancy, and one month following pregnancy end date, or censoring (due to loss to follow-up, death), whichever occurs first.
Data sources
The following data sources, which have already implemented the PET, will be used for this study.
1. Sistema d’Informació per al Desenvolupament de la Investigació en Atenció Primària (SIDIAP), Spain.
2. The Norwegian Linked Health Registry data (NLHR), Norway.
Statistical analysis
Characterisation of users of benzodiazepines and alternative treatments during pregnancy and individuals experiencing pregnancy losses will be done using PatientProfiles and CohortCharacteristics R package. For objectives 2 and 4, we will use the DrugUtilisation and TreatmentPatterns R packages to characterise benzodiazepines and alternative treatments during pregnancy, including counts (%) for each class, duration of treatment, prior medication use, and trajectories of prescription fill. For the calculation of the prevalence of benzodiazepines and alternative treatments during pregnancy, and the incidence rates of the outcomes of interest, the IncidencePrevalence R package will be used. Rates will be reported with the 95% Poisson confidence intervals. A minimum cell counts of 5 will be used when reporting results, with any smaller count reported as “<5”. All analyses will be reported by country/database, overall and stratified by age groups and pregnancy period, when possible (minimum cell count reached). No meta-analysis will be performed.