Core Data Elements for Pregnancy Pharmacovigilance Studies Using Primary Source Data Collection Methods: Recommendations from the IMI ConcePTION Project

The CDE recommendations are provided in 14 tables of related data elements. The first peer-reviewed version of these tables is provided in the supplementary appendix (see the electronic supplementary material). Live versions of the recommendations are also published in full on the ENTIS website (http://www.entis-org.eu/cde). These recommendations will be updated when required to ensure the data elements and their definitions remain aligned with current PregPV research practices and needs. Below we provide a brief overview for each of the CDE tables and highlight important content/considerations.

3.1 Database Administrative Details

Online Table 1 describes data elements that would enable administrative tasks and/or database functionality when using primary source data collection methods in PregPV research.

PregPV research often focusses on prospectively reported pregnancies to limit the impact of inclusion or sampling biases. The prospective and retrospective definitions in Online Table 1 resulted in considerable discussion and deliberation because there are multiple, sometimes contradictory, definitions of a prospective pregnancy report in both the literature and guidance documents from regulatory authorities. Published definitions can also be ambiguous and are potentially open to differing interpretations. For example, the FDA define a prospective pregnancy report as being described ‘prior to any information about the pregnancy outcome being available’. This may be interpreted to include all pregnancies reported prior to the end of pregnancy if ‘outcome’ refers to birth, miscarriage, or termination, but is ambiguous in the content of congenital anomalies as a defined outcome. Through discussion within the CDE working group, the FDA definition was interpreted as any pregnancy reported prior to prenatal screening having been undertaken (to detect congenital anomaly). However, the EMA definition describes a prospective report as being described ‘prior to the knowledge of the pregnancy outcome or prior to the detection of a congenital anomaly at prenatal examination’. This was interpreted slightly differently as any pregnancy reported prior to any prenatal detection of congenital anomaly (and thus could include pregnancies where prenatal screening had been undertaken, but where no abnormalities were detected). The CDE recommendations are also applicable to retrospectively collected data.

For this reason, the definition of prospective status provided in the CDE recommendations is the most basic; pregnancies reported to the data collection system whilst the patient is still pregnant. This definition was considered preferential as it was thought to be the least restrictive upon study sample sizes. Furthermore, this definition can be applied to studies investigating postnatal childhood health and neurodevelopmental outcomes, recognising that the outcomes remain unknown for the entire pregnancy. However, the recommendations also highlight that alternative definitions for a prospective pregnancy report exist, and that these may be used where preferred or required, and may vary by the outcomes being studied. These definitions allow for the exclusion of pregnancies that have been recruited either (1) after any prenatal screening has been performed [33] or (2) after the prenatal detection of congenital anomalies or any other adverse pregnancy/foetal outcome [34]. The CDE recommendations therefore also describe additional data elements that are of value to collect, in order to apply other commonly proposed definitions of a prospective pregnancy case.

3.2 Maternal/Paternal Details

Online Table 2 provides recommendations for important co-variable risk factors relating to the pregnant woman and the father of the conceptus. These include sociodemographic factors such as the pregnant woman’s age, income, educational attainments, ethnicity, body mass index (BMI), and details about smoking, alcohol, and recreational drug use, among others, all of which were judged to potentially impact on pregnancy/foetal/neonatal outcomes. Whilst many of these co-variables apply to a range of such outcomes, certain co-variables may pertain only to longer-term child health or neurodevelopmental outcomes. Many data elements included in Online Table 2 are considered important co-variable risk factors, and collection of these data is highly recommended. For some of the data elements, such as maternal ethnicity, there is a limited understanding of the influence these variables have on pregnancy outcomes. Inclusion of these variables in the CDE recommendations may result in more research to understand these interactions. Additionally, some of the data elements, such as maternal IQ generated from a clinically validated tool, may not be readily available. Where feasible, research tools and infrastructure may need to be developed to allow for the collection of information that is considered of high importance for the specific outcome under investigation. This may be particularly relevant to child health and neurodevelopmental outcomes.

3.3 Pregnancy Details

The CDE items listed in this Online Table 3 are valuable for data collection administrative duties, such as identifying when pregnancy outcome follow-up attempts should be performed, for deriving alternative prospective/retrospective definitions, and certain pregnancy outcomes. Specifically, the table contains the expected date of delivery (EDD), which is a vital data item for PregPV purposes. This data item provides an anchoring point for several essential CDE items, including the trimester of pregnancy at set time points and exposure time periods (when reported by date only), and is informative for pregnancy outcomes such as gestational age at delivery (prematurity) and expected birth weight (small or large for gestational age). In the CDE, the pregnancy monitoring period is deemed to start from the date of last menstrual period (LMP) where this is available. In most pregnancies, this is approximately 2 weeks prior to the approximate date of conception. For PregPV studies utilising primary source data collection, the LMP is a useful date of reference for data collection purposes. It also allows estimation of gestational age at set time points early in gestation, before a more accurate dating of the pregnancy has been established through ultrasound dating scans. Alternative methods to calculate gestation are described for instances where the LMP is not available or is unreliable (for example, due to an irregular menstrual cycle).

3.4 Maternal Medical History Details

Online Table 4 contains a single CDE item that provides recommendations about collecting details of maternal medical conditions that existed prior to pregnancy. Of note, the details collected here can reflect the indication(s) for medication use (details of which are collected in Online Table 6), both for any medication/medicinal product under investigation and for any concomitant medications. In many cases, it would be valuable to collect information about the severity of the condition(s), particularly whether these conditions are active at the time of conception. Recommendations regarding essential variables for specific maternal diseases were beyond the scope of the ConcePTION project. Investigators are therefore strongly encouraged to liaise with maternal disease specialists to develop disease-specific variables that may correlate with adverse pregnancy outcome or impaired child health and/or neurodevelopment. Maternal illness may be an important co-variable risk factor for the analysis (at either the individual case level or in statistical covariate analysis) of certain adverse pregnancy/foetal/neonatal/infant outcomes. Details regarding maternal mental health conditions are considered particularly valuable for studies investigating longer-term child health and neurodevelopmental outcomes.

3.5 Family Medical History and Obstetric History Details

The items described in Online Table 5 relate to relevant family medical history, including those of genetic conditions, congenital anomalies, neurodevelopmental impairments, and obstetric history, including the number of prior pregnancies and their outcomes. Such information may be important as co-variable risk factor(s) for the analysis (at either the individual case level or in statistical covariate analysis) of certain adverse pregnancy/foetal/neonatal/infant outcomes.

3.6 Pregnancy Medication Exposure Details

Collection of high-quality data around the timing, dose, route of administration, and indication for medication use in pregnancy is crucial when performing PregPV research. Each of the CDE items listed in Online Table 6 should be repeated for each instance of medication use during pregnancy. An instance of medication use is a time period when a medication was used at regular intervals. Should an individual medication be stopped and then restarted later in pregnancy, it is recommended that this is logged as a separate instance of use.

Several of the CDE items in this table required significant discussion and debate, particularly for the definitions provided for the timing of peri-LMP exposures and the trimesters of pregnancy. The first trimester has various definitions in the literature, both in terms of when the period begins and also when it ends. Some PregPV studies state that the first trimester period starts at the date of LMP, whereas others define the first trimester period begins from conception (approximately 2 weeks post-LMP). This minor discrepancy can impact on exposure classification and hence adverse outcome risk estimates.

Similarly, there is no standard definition for the period before the pregnancy began during which effects of exposure to medications may persist into pregnancy (here referred to as the peri-LMP period). Researchers may need to adapt the definition of the peri-LMP period (see Online Table 6) at the analysis stage depending on the medication exposure under study, taking consideration of the pharmacokinetics and pharmacodynamics of a particular medication. For example, medications with long half-lives may be discontinued prior to conception, but clinically relevant concentrations of the medication may remain in the systemic circulation into the sensitive period of organogenesis. Alternatively, medications that are used to treat acute conditions may only be used for short time periods and be discontinued between the LMP date and the date of conception. Provided these medications have a short half-life, pre-conception/peri-LMP exposures probably have limited relevance with regard to malformation risks in the developing foetus.

3.7 Maternal Illnesses and Obstetric Complications Details

Online Table 7 describes items that relate to medical conditions arising during pregnancy and are important co-variable risk factors for assessing aetiological relationships between gestational medication use and adverse pregnancy/foetal/infant outcomes. Some of these items may themselves also be important outcome variables for certain medications, for example, the occurrence of gestational diabetes in a study of glucose-lowering medications. The aim for these items is to collect information on any new maternal illnesses that develop during the pregnancy, including conditions related to pregnancy such as gestational diabetes, pre-eclampsia, or nausea and vomiting in pregnancy/hyperemesis gravidarum and non-pregnancy specific conditions. As shown by the recommended data format, it is strongly encouraged that data are collected on the gestational age when conditions arise. Collection of this detail will allow researchers to analyse the impact of maternal complications/illnesses at aetiologically relevant time points in the pregnancy.

3.8 Pregnancy Outcome Details

Collection of high-quality data about pregnancy outcome is vital. Online Table 8 describes the recommended data elements relating to pregnancy outcomes. The data items spontaneous abortion (SA) (miscarriage) and stillbirth (intrauterine foetal death) required extensive debate. Definitions vary in the literature for the gestational age when a foetal demise is termed a miscarriage or an intrauterine death. This likely stems from international variation in definitions about when a foetus reaches a stage (based on gestational age alone or in combination with estimated foetal weight) where preterm birth could result in survival. Definitions in the literature range from 20 to 28 weeks gestational age (post-LMP) as the upper limit of when a foetal demise would be considered an SA/miscarriage [46,47,48]. Some definitions require the foetus to have reached a weight of at least 500 g before the demise is defined as an intrauterine foetal death [49]. After extensive discussion within the CDE recommendations working group, and externally with expert members of the ConcePTION project, the definition for SA/miscarriage was determined to be a foetal demise before the start of the 22nd week of pregnancy (≤ 153 days post LMP), with any foetal demise after this point being defined as a stillbirth (intrauterine foetal death). As stated previously, collection of gestational age at foetal demise is strongly encouraged as it allows outcome definition standardisation within the dataset, as well as the flexibility to undertake sensitivity analyses using different definitions if required. Data from datasets that have used varying definitions for SA/miscarriage and stillbirth could also be standardised for combination in common data models.

3.9 Delivery Details

Online Table 9 describes the recommendations for information regarding delivery details. These data elements may provide supplemental data of relevance for the analysis of certain adverse pregnancy outcome events, such as perinatal death or neonatal complications. It is recommended that the CDE items listed in this table are collected for each delivery event in the reported pregnancy. On rare occasions in multiple-foetus pregnancies, there may be more than one delivery event at different times and delivery methods may vary for each foetus/infant.

3.10 Live/Stillborn Birth Outcome Details

The CDE items listed in Online Table 10 should be collected for each live or stillborn infant.

3.11 Live Born Neonatal/Infant Outcome Details

The CDE items listed in Online Table 11 should be collected for each live born infant. The data item ‘Product/disease-specific outcomes’ included in this table is deliberately non-restrictive to allow researchers to customise their data collection efforts to gather information on relevant postnatal outcomes, defined by the medication and/or the primary outcome under investigation. For example, researchers investigating immunosuppressive medication use in pregnancy may wish to expand on the CDE recommendations to collect information about postnatal immune function, such as the number of infections experienced over a set period of time. Details relating to neonatal/infant death are also included in this table.

3.12 Malformations Details

The CDE items listed in Online Table 12 should be collected for each infant or foetus from the reported pregnancies. These details may not be available for foetuses that have been terminated due to non-medical reasons and are unlikely to be available for pregnancies that resulted in an SA/miscarriage. Where foetuses/infants are identified as having congenital anomalies, the CDE recommendations advise that full details of each anomaly are collected. The CDE recommendations currently endorse the use of EUROCAT guidelines to categorise each anomaly present in the foetus/infant. The CDE also makes provision for the recording of the classification of the case (as opposed to classification of individual anomalies). This section should only be completed following expert committee review, preferably whilst blinded to exposure status. Currently proposed categories include genetic/cytogenetic anomalies, major malformation, minor malformation, or malformations not otherwise stated (NOS). More details about classification are provided in the table. It is recommended that each congenital malformation event, and the resulting classification of the infant or foetus, should be judged by experienced adjudicators (qualified paediatrician, clinical geneticist, teratologist, paediatric neurologist, nephrologist, toxicologist, or clinical pharmacologist).

3.13 Longer-Term Child Health and Neurodevelopmental Outcome Details

Many PregPV systems are not structured to investigate longer-term child health or neurodevelopmental outcomes. Recently, experts within the field have recommended the inclusion of child neurodevelopment outcomes more centrally, to reduce the latency between a medication’s approval and the development of conclusive evidence regarding neurodevelopmental risk [22].

Child health and neurodevelopmental outcomes are numerous, can vary over time, and can be investigated to different levels of sensitivity. Longitudinal follow-up is recommended due to the prolonged period of development. It is recognised that a single study or initiative may not be able to investigate all outcomes to a high level, but it is recommended that as comprehensive a range as possible be included in all initiatives.

The data items included in Online Table 13 relate to childhood health outcomes, whilst Online Table 14 describes the data items recommended for assessing neurodevelopmental outcomes. The recommended child health questions have been devised from a list of common child health difficulties [50] and are considered sufficient to provide a high-level overview of childhood health. Child health is very complex, and numerous conditions may present. For the purposes of the CDE recommendations, it was not considered practical to include all of these possible conditions as essential elements. If researchers are aware of any specific conditions related to either an exposure or outcome of interest, these should also be included specifically. There may also be benefit in including detail on condition severity.

Neurodevelopmental outcomes represent the development and functioning of the central nervous system and are a set of independent, but interlinked, skills that evolve over time. Given the variety of PregPV data collection initiatives that may wish to adopt the CDE recommendations, only high-level neurodevelopmental domains have been included in Online Table 14. Measurement of these functions will vary by study design, investigator expertise, and the age of the children at assessment. Expert consultation should be undertaken to design the appropriate investigations.

It is recognised that not all child longer-term outcomes will be covered by the data items in these tables; however, the listed elements can be complimented by more detailed and in-depth studies investigating specific developmental attainments and health endpoints/outcomes.

留言 (0)

沒有登入
gif