Skip to main content
Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Where are the data linking infant outcomes, breastfeeding and medicine exposure? A systematic scoping review

  • Sue Jordan ,

    Roles Conceptualization, Data curation, Funding acquisition, Investigation, Methodology, Writing – original draft, Writing – review & editing

    Affiliation Faculty of Health and Life Science, Swansea University, Swansea, Wales, United Kingdom

  • Sophia Komninou,

    Roles Data curation, Formal analysis, Investigation, Methodology, Writing – review & editing

    Affiliation Faculty of Health and Life Science, Swansea University, Swansea, Wales, United Kingdom

  • Sandra Lopez Leon

    Roles Formal analysis, Funding acquisition, Investigation, Methodology, Writing – review & editing

    Affiliations Quantitative Safety & Epidemiology, Novartis Pharmaceuticals, East Hanover, NJ, United States of America, Rutgers Center for Pharmacoepidemiology and Treatment Science, Rutgers University, New Brunswick, NJ, United States of America



Information on the impact of medicines on breastfeeding and the breastfed infant remains scarce. The aims of this review were to identify databases and cohorts holding this information, and pinpoint current information and research deficits.


We searched 12 electronic databases, including PubMed/ Medline and Scopus, using a combination of controlled vocabulary (MeSH terms) and free text terms. We included studies reporting data from databases with information on breastfeeding, medicines exposure, and infant outcomes. We excluded studies not reporting all three parameters. Two reviewers independently selected papers and extracted data using a standardised spreadsheet. Risk of bias was assessed. Recruited cohorts with relevant information were tabulated separately. Discrepancies were resolved by discussion.


From 752 unique records, 69 studies were identified for full review. Eleven papers reported analyses from ten established databases with information on maternal prescription or non-prescription drugs, breastfeeding and infant outcomes. Twenty-four cohort studies were also identified. No studies reported educational or long-term developmental outcomes. The data are too sparse to warrant any firm conclusions, beyond the need for more data. The overall picture hints at 1) unquantifiable, but probably rare, serious harms to infants exposed to medicines via breastmilk, 2) unknown long-term harms, and 3) a more insidious but more pervasive harm in terms of reduced breastfeeding rates following medicines exposure in late pregnancy and peri-partum.


Analyses of databases reporting on the full population are needed to quantify any adverse effects of medicines and identify dyads at risk of harm from prescribed medicines while breastfeeding. This information is essential to ensure 1) infants are monitored appropriately for any adverse drug reactions 2) inform breastfeeding patients using long-term medicines as to whether the benefits of breastfeeding outweigh exposure to medicines via breastmilk and 3) target additional support to breastfeeding patients whose medicines may affect breastfeeding. The protocol is registered with the Registry of Systematic Reviews, no.994.


Establishing health service databases and databanks has been costly in time, energy, and money. Their contribution to pharmacovigilance is considerable, particularly where randomised controlled trials are impossible for ethical and logistical reasons, for example during pregnancy and lactation, and where outcomes are so rare that impossibly large numbers of people would need to be recruited to demonstrate statistically significant differences for such outcomes (for example, many congenital anomalies). However, the value of health service databases holding electronic records of routine care and observational research is limited by the data collected, both the variables recorded and their completeness. Population databases provide insights into the determinants of health and the impact of medicines in pregnancy on infant outcomes, but only five European databases have any data on breastfeeding: the national databanks for Finland, Scotland and Wales, EFEMERIS / POMME in Haute-Garonne, and hospital records of breastfeeding at discharge in Tuscany [1].

Breastfeeding is complex, with nutritional, immunological, and psychosocial aspects, which are not easily disentangled. It profoundly affects women and children. Benefits to infants include reduced: mortality (particularly necrotising enterocolitis and sudden infant death syndrome), gastro-intestinal and respiratory infections, acute otitis media, asthma/wheezing [2, 3], malocclusion, obesity and type 2 diabetes. Benefits to mothers include reduced rates of breast and ovarian cancers, type 2 diabetes, myocardial infarction and hypertension [2]. In the USA 3,340 (95% confidence interval 1,886 to 4,785) annual excess deaths are attributed to shortened duration of breastfeeding (defined as less than 1 year, and exclusively <6 months): 78% of these excess deaths are maternal, and 22% infant [2].

The safety of a medicinal product during lactation is complex, in that it involves the effects of medicines on both infant and mother plus the interactions and bonding between them. Their very different pharmacokinetics (particularly elimination half-lives), and the need to calculate these for mother, neonate, and preterm neonate complicate determination of safety. Before a medicine’s safety profile can be considered complete, several questions need to be addressed:

  • How does the medicine affect the physiology of lactation?
  • How are breastfeeding rates affected by administration during pregnancy, labour, the puerperium and during lactation?
  • Can these effects be mitigated by recognition, support, and clinical management?

What is the effect of the medicine on the breastfed infant? Some 70% infant ADRs are dose-dependent [4], but concerns remain regarding preterm infants and those with allelic variations in key enzymes–the extreme phenotypes [1].

  • How should the infant be monitored for any possible adverse effects?
  • Do the benefits of breastfeeding outweigh possible disbenefits from exposure to medicines via breastmilk, short- and long-term?

Currently, studies reporting breastfeeding, its predictors and consequences are, with few exceptions, based on recruited cohorts [5]. Some existing cohorts with potential for pharmacovigilance, such as the Millennium Cohort Study [6], the Norwegian and Danish mother and baby studies [7, 8], are linked with population databases. Without full population coverage it will be difficult to report associations free of volunteer [9], and collider bias. These arise when samples do not represent the population, because volunteering is related to variables being investigated, such as medicines exposure, breastfeeding or social class [1014].

Although population databases are an important advance in pharmacovigilance, it appears that they may be less than comprehensive, particularly for issues affecting women and children, including pregnancy prevention programmes aiming to reduce exposure to known teratogenic medicines [15]. If pregnant and breastfeeding women and children are not to be excluded from global pharmacovigilance initiatives, population databases with information on breastfeeding for the full population should be identified. We defined a database as “a structured set of data held in computer storage” [16], more specifically, a large collection of data organized and maintained so that it can be expanded, updated, and retrieved rapidly for various uses [17]. To inform the discussion around implications for practice, information from cohort studies [18], not derived from databases, was tabulated. This systematic scoping review aimed to identify and report the databases and cohorts with information on breastfeeding and its impact on infants, and summarise any apparent information and research deficits.


We conducted a scoping review using systematic searches to map and locate the databases providing quantitative evidence on medicines exposure, breastfeeding and infant outcomes, and summarise the evidence [19, 20].

The protocol for this search is registered [21] (S1 File). This review follows the PRISMA guidelines [22], and the extension for scoping reviews [23].

Search strategy

Twelve electronic databases (PubMed/Medline, Scopus, CINAHL, PsycINFO, Web of Science, British Nursing Database, Proquest, Drugs and Lactation Database (LactMed), ZETOC, TRIP, MIDIRS, Wiley Online Library) were searched to May 2022 using a combination of controlled vocabulary (MeSH) and free text terms. These included terms for breastfeeding, lactation, or infant feeding along with terms for pharmacovigilance or drug monitoring or drug surveillance. The search strategy is shown below. There were no language or date or location restrictions, but the search was restricted to papers reporting on humans only.

Search strategy.

Search terms. “Breastfeeding OR Lactation OR Breastfe* OR Breast-fe* OR “Breast fe*” OR Lactat* OR “Infant feed*” OR “Infant Nutrition”


“Pharmacovigilance OR Product Surveillance OR Postmarketing OR Drug Monitoring OR Adverse Drug Reactions OR Pharmacovigilan* OR “Drug monitor*” OR “Postmarketing Surveillance” OR “Post-marketing Surveillance” OR “Post marketing Surveillance” OR “Adverse Drug Reaction*”


For two databases (PubMed and PsychINFO) it was necessary to specify “NOT economics” in order to obtain more relevant results.

Inclusion and exclusion criteria

Inclusion criteria.

Reports from databases or cohorts with empirical data on breastfeeding plus human maternal medication exposure plus infant outcomes plus pharmacovigilance or adverse drug reactions.

Exclusion criteria.

  • No empirical data on infant feeding/breastfeeding, maternal medicine exposure and infant outcomes / welfare.
  • Single case reports.
  • Cross sectional surveys
  • Articles not in English with neither an English abstract nor empirical data tables.

We excluded pharmacokinetic plasma / milk transfer studies and case series where there was no information on infant outcomes. Where infant outcomes were reported, we included these cohorts. Literature reviews were excluded but reference lists were examined for further databases. We excluded papers a) without empirical data and b) not reporting infant outcomes. Tabulation aimed to describe the database or cohort (size, location) and participants, exposures or interventions (medicines, doses and timing), outcomes (breastfeeding rates and infant welfare or ADRs), and the inferences of the investigators. After initial data extraction, some items were collapsed where there was a paucity of information, for example on comparators and long-term outcomes.

Study selection

Following the search, duplicates were removed, and publications were screened by titles to identify those likely to meet the study inclusion criteria. This was carried out independently by two blinded researchers (SJ/SK or SLL/SK). The titles and abstracts or first pages of the remaining studies were reviewed by two researchers, blinded (SJ/SK or SK/SL) according to inclusion and exclusion criteria. Papers were then selected for full review. Full texts of all articles selected for consideration were retrieved, read, and decisions on inclusion were reached jointly. The reference lists of included studies were reviewed to identify other possibly relevant studies. These studies were then reviewed following the same process outlined above. We separated the studies reporting established databases from those reporting recruited cohorts. The relevant details from included papers were tabulated and checked independently (by SJ and SK and re-checked by SLL) (Tables 1 and S1).

Table 1. Studies using databases reporting medicines use during breastfeeding (chronological order).

Tabulated information was summarised, in accordance with the review’s objectives to describe the databases reporting on medicines, breastfeeding and infant outcomes simultaneously, and report the purported effects of medicine exposure [20]. A critical appraisal of the risk of bias in the database reports was based on a recognised tool for assessment of non-randomised studies of exposures using consensus-driven domains relating to: confounding, selection, intervention misclassification or mismeasurement, post-exposure interventions, missing data, measurement and selective reporting (ROBINS-E) [24]. We were unable to assess the direction of bias (SJ, checked by SK, SLL).


Searches identified 858 titles. A further four studies were identified by reviewing the reference lists of included studies, a total of 862. Removing duplicates reduced numbers to 752. We were unable to identify an abstract for 52 of these, so they were reviewed by title, date and provenance. First page or pdf was identified for 35. Seventeen were book chapters, 11 were editorials, 7 were clinical notes. Seventeen, had neither first page nor abstract. Ten studies, dated from 1966 to 1999, had no email contact details. We contacted the remaining seven authors but received no responses.

Review of titles and, if needed, abstracts or first pages of the remaining 700 studies identified 69 papers for full review. The most common reasons for exclusion were: ‘out of scope’, absence of empirical data (mainly reviews), and absence of data on infant outcomes or welfare. We excluded 14 papers that could not be retrieved in full and pre-dated 2004, the year of the earliest database identified in our earlier work [1]. 36 papers had neither an English abstract nor empirical data tables. Of the 69 papers initially identified for further review, 33 were excluded as they did not meet the inclusion criteria (for detailed exclusion reasons, see PRISMA diagram, Fig 1), leaving 36 studies (Fig 1). Most (18) excluded papers did not contain data from databases or cohorts, seven described the transfer of medicines into breastmilk but did not report on infant outcomes, even to say that infants were well. Three database studies described purchase of medicines, and one infant medication. The paper on the negative impact of pesticides on breastfeeding rates [25] was considered ‘out of scope’. Of the two papers with information on dose-response one was a single case report [26], and excluded; the other report emanated from a cohort of 7 infants (S1 Table) [27] All papers with more than one participant with relevant data were tabulated. Details of all included database studies were extracted including study objectives, study location, and details of exposures, participants, outcomes, and findings (Table 1).

Databases identified

Ten databases were identified (Table 1). We included the Norwegian MoBa cohort in this table, as it is linked to the database of Norwegian national records of medicines dispensed in primary care. However, data on breastfeeding were only available for patients linked with a volunteer prospective cohort study [28]. The PRAMS database contains whole-population data on infant outcomes, but marijuana (cannabis) use was taken from self-reported questionnaires from a stratified sample of live births across the USA and links with birth certificate information [29]. Spontaneous reports formed the basis of five studies and four databases [3034]: two of these studies were from the olanzapine manufacturer’s databases, and one from glatiramer acetate manufacturer’s database [33]. We identified only one national database reporting adverse drug reactions (ADRs) in breastfed infants: the French spontaneous reporting database [30], and only one report of an adverse event affecting breastfeeding in the Uppsala Monitoring Centre international database [32]. Four medicine information centres generated databases based on patients’ spontaneous contacts followed up by telephone to ascertain outcomes: two in Israel [35, 36], and one each in California [37], and Canada [38]. The Centre for Disease Control’s (CDC’s) Pregnancy Risk Assessment Monitoring System (PRAMS) was used to report data on recreational drugs in two papers [29, 39].

Three papers emanated from olanzapine surveillance: two from manufacturers [31, 34], and one from an information service [36]. Two papers from one database reported on recreational drugs [29, 39]. Four reported on specific drug groups: psychotropics (any) [35], SSRI antidepressants [37], antiepileptics [28], monoclonal antibodies for migraine [32]. Only two papers reported use of any or all medicines [30, 38]. Three compared exposures to psychotropics with other medicines: antibiotics [35], paracetamol or dental treatment [37], paracetamol [36]. [28] compared outcomes for those using AEDs with unmedicated epilepsy and the reference populations. [29, 39] compared outcomes for those using or not using marijuana. Nine papers reported on both pregnancy and breastfeeding exposures [29, 3133, 3538]. Psychotropic medicines, but not marijuana, appeared to be associated with a relatively high incidence of suboptimal perinatal outcomes, including withdrawal reactions and poor suckling [36].

Six databases reported on the impact of prescription medicines on breastfeeding [28, 29, 31, 3538]. No databases related to hospital prescribing, although spontaneous reports did not specify the provenance of prescriptions. None held information on medicines in labour.

From ten papers, data on 4,264 dyads exposed to prescription medicines were reported, and two further papers reported on 8,176 dyads surveyed regarding marijuana (cannabis) [29, 39] (total 12,440).

Cohorts identified

We identified 24 papers reporting recruited cohorts, each describing a single study. These are presented in S1 Table. Cohorts reporting infant outcomes, maternal medicines exposures and breastfeeding ranged in size from three to 1719 infants: seven included less than 10 infants. One large cohort represented follow up from two randomised controlled trials (RCTs) in Botswana: the cumulative incidence of severe anaemia in breastfed infants varied between antiretroviral regimens [40]. Nine cohorts reported on breastfeeding as an outcome, and all 24 on infant outcomes. No studies reported using breastfeeding as a covariate.

Impact on breastfeeding rates

In cohorts and databases, prescription medicines, marijuana and pesticides adversely affected breastfeeding rates. Reasons for shortened breastfeeding duration included: patients’ concerns over prescription medicines [38, 41], weak suckling [42] or adverse effects on the infant [43, 44]. Decreased lactation following oestrogen and progesterone exposure led to early discontinuation of breastfeeding [38, 45]. Early discontinuation of breastfeeding was also associated with mental health medicines [35], including olanzapine [36], SSRIs [37], antiepileptics [28, 42].

Impact on infants

The French pharmacovigilance databases provided whole-population data, but relied on spontaneous reports, which may underestimate ADR prevalence by over 90% (Hazell & Shakir 2006). Selection, volunteer, and collider bias were not reported in any papers. Medicine exposure via breastmilk affected some, not all, infants. Two databases and one cohort reported that some infants exposed via breastmilk experienced serious ADRs, mainly the known adverse effects of medicines [30, 36, 40]. For example, following exposure via breastmilk there were cases of: infant apnoea following maternal use of benzodiazepines or opioids: haemorrhage and infant renal insufficiency following ketoprofen; and neutropoenia following carbimazole [30]. There were single case reports of hypotension associated with a beta blocker [46] and impaired suckling and vomiting with carbamazepine [42].

Some, but not all, infants whose mothers took benzodiazepines or opioids [30, 38, 44, 47], olanzapine [34] or other mental health medicines [35, 48], were sedated or sleepy or constipated [48], which may have led to failure to gain weight due to insufficient feeding [36]. Infants exposed to SSRIs were more likely to be irritable and/ or feeding poorly [43, 49], but this was not reported in all studies [37], particularly the small cohorts [4951]. Dose were not always reported, and there were no dose-response analyses. The only exploration of the effect of dose was a report of olanzapine dose reduction resolving drowsiness for one infant [27]. Similarly, reducing the dose of citalopram improved infant sleep in a separate single case report [26].

‘Minor’, well-known adverse effects were pervasive, affecting 94 out of 838 infants [52]: these included infant diarrhoea following maternal antibiotics [35, 52] or antipsychotics [48], and infant oral Candidiasis following metronidazole [53]. Two papers from one database reported no adverse effects in infants exposed to marijuana [29, 39].

There were no reports of educational outcomes or follow up beyond 3 years. Twelve studies reported various developmental outcomes [28, 35, 36, 49, 50], including six small cohorts (with <11 participants) [27, 51, 5457]. One of six infants [27], three of 22 [36] exposed to olanzapine, five of 28 exposed to olanzapine, risperidone or quetiapine [48], and one of 10 exposed to antidepressants [54] exhibited developmental delay. Following in utero exposure to antiepileptics, breastfed infants were less likely to exhibit autistic traits than formula fed infants [28]. Four of 10 infants exposed to lithium had abnormal results for renal or thyroid function in venous blood samples, but no other observable ADRs; long-term sequelae were not ascertained [56].

Risk of bias

Most analyses were descriptive, and based on biological plausibility. Few analyses accounted for all known confounding variables, such as socio-economic status (SES), alcohol use [58], or pesticide exposure [25], despite known associations. The impact of fluctuations in milk composition and fat content or the potential for increased exposure associated with clinical or subclinical mastitis were not discussed [59, 60]. No studies defined the extent of breastfeeding, whether exclusive or partial: recent studies relied on self-report [33, 48, 61]. All studies relied on maternal self-report of breastfeeding: this may over-estimate duration [62] and initiation of breastfeeding [63], but is considered reasonably accurate if recalled within 3 years [64].

Only one study [40] involved randomisation: this is likely the definitive work on exposure to highly active anti-retroviral therapy (HAART) regimens, but it did not involve a database. It was the only cohort study with >200 participants. The six databases from information services [32, 33, 3538] were vulnerable to bias emanating from self-selection and non-response to follow up. The PRAMS database was vulnerable to non-response bias, despite exhaustive attempts at telephone contact, and the MoBa recruited cohort to volunteer bias. The databases relying on spontaneous reports [30, 34], were crucial in signal generation, but may fail to identify the majority of ADRs.

Although the database studies were well-conducted, these inherent limitations in their design puts them at moderate risk of bias, at best (Table 2).

Table 2. Appraisal of risk of bias in databases identified (chronological order).


The data available from databases and cohorts are too sparse to justify any firm conclusions, beyond the need for more data. Similarly, a scoping review of post-marketing studies identified only 10 studies reporting infant exposure during breastfeeding [65]. The absence of data from whole-population databases compounds concerns:

  • Serious inter-generational ADRs from exposure via breastmilk are unquantifiable, but appear to be rare; however, concerns remain, particularly for medicines acting on the central nervous system (CNS).
  • There is sufficient information to warrant frequent detailed monitoring of infants exposed via breastmilk, above and beyond routine ‘well-baby’ checks. There are no data to indicate that infant monitoring is unnecessary.
  • For some medicines, there is insufficient information to advise patients whether the benefits of breastfeeding outweigh the harms from exposure via breastmilk.
  • The more insidious but more pervasive harm of reduced breastfeeding rates following medicines exposure in late pregnancy, labour and peri-partum will remain unquantifiable until whole-population database and hospital prescribing pharmaco-epidemiological studies are undertaken.

The ADRs and harms to infants identified here reflect those reported in comprehensive reviews [66] of case series, small cohorts, databases [4, 67], and manufacturers’ literature. Two of 10 databases identified focused on olanzapine: two papers from the same database were sponsored by the manufacturers. The later paper, [34] is more reassuring than the earlier paper [31], but indicates that exposure during breastfeeding adversely affects 15.6% (16/102) infants, without reporting recovery or long-term outcomes. Concerns are supported and extended to risperidone and quetiapine in a small cohort [48]. This underlies the firm advice of manufacturers and the British National Formulary [68]. We identified very little data for alternative second-generation antipsychotics, and none for aripiprazole, where the BNF states ‘manufacturer advises avoid’, rather than simply ‘avoid–present in milk’ (no.83 p.430).

The benefits of breastfeeding to the infant, despite exposure to medicines for epilepsy via breastmilk, were apparent in the cohort with detailed long-term follow up [28], and other observational work; however, more data are needed for some AEDs, particularly phenytoin [69], ethosuximide, phenobarbital or primidone [70]. Further exploration is essential to review the benefit/harm balance, as other databases report only ‘gross motor development’ [35], rather than the full range of outcomes.

We have little information as to why breastfeeding rates were lower amongst those using prescription medicines or why people discontinued breastfeeding. However, the lower rates are consistent with those reported elsewhere [71, 72], and may be influenced by the absence of large studies offering reassurance of safety [73, 74] or the serious adverse effects reported in a small number of infants [30].

Wider implications: The information desert

This review, like others [73], identified that exposures to medicines were associated with reduced breastfeeding rates. It also indicated that other exposures, such as recreational drugs [29, 39], may have a similar effect, suggesting that these exposures should be accounted in observational studies.

Currently, manufacturers are not obliged to provide data on breastfeeding. Data from animal studies are of uncertain value, as milk composition, and hence drug transfer, differ between species [60, 75]. The U.S. Food and Drug Administration (FDA) asks manufacturers to provide data on the impact of medicines on breastfeeding, and the breastfed infant, but this is not mandatory and there is an option for the ‘lactation’ section of product information to be omitted [76]. Current Medicines and Healthcare products Regulatory Agency (MHRA) guidance on UK product labelling in lactation indicates: ‘If available, clinical data from exposed breastfed infants should be mentioned as the conclusions of kinetic studies’ ([77] p.11). The European Medicines Agency (EMA) suggests that studies on breastfeeding ‘could be considered’, whilst noting, as indicated in this review, that ‘Reliable information regarding patient exposure in breastfeeding is not routinely available but may exist in some European birth cohorts.’ ([78] p.22). Alongside calls for further pharmacokinetic and pharmacogenomic studies [75], the concerns raised by the existing databases and cohorts should stimulate change in the availability of full-population databases with breastfeeding data [1, 79] (Yalcin et al 2022).

Limitations of the data

Signals were generated by the studies in this review, but not pursued. Most authors based the associations reported on biological plausibility [80], rather than effect estimates: the corpus of literature supports the supposition that some infants may be vulnerable to the known ADRs of medicines transferred via breastmilk, but to an unknown and unpredictable extent. There was little information on dyads: most ADRs were reported in term infants or without specified gestation. The reduced renal function and impaired drug clearance in preterm infants [81, 82] suggests that omission of this vulnerable group may lead to under-reporting of harm.

No assessments of data quality were provided, and these are reported to be generally lacking even in large databases [83, 84]. Only one cohort [40] and six databases [28, 29, 34, 35, 37, 38] had >200 infants with breastfeeding data, the minimum sample size to detect serious adverse events in neonates [85]. It is estimated that spontaneous reports identify some 5% of ADRs [86], and the”less serious” more insidious reactions are particularly vulnerable to under-reporting [87]. This suggests that a more comprehensive approach to data collection is needed than provided in existing databases [30, 31, 34]. We have no indication as to the impact of any recall bias, volunteer selection bias, or social desirability response bias. These may over-estimate the prevalence of breastfeeding and under-estimate harms, which are over-represented in the most disadvantaged sections of the population [71, 88].

Like all non-randomised studies, those identified were vulnerable to unmeasured confounding, including unknown or lurking variables [89], and confounding by indication [90]. Selection, volunteer and collider bias impacts on studies that are not ‘whole-population’ [1, 12], including the databases identified here (Table 2). Their findings cannot be automatically transferred to the sections of the population who did not participate, mainly the economically disadvantaged [9], and recruitment by self-selection can distort associations via collider bias [12]. For example, when exploring the impact of medicines on initiation or duration of breastfeeding, if recruitment to the database or cohort were to favour participants who were both using medicines and breastfeeding, these characteristics would be over-represented. This over-representation would distort the sample and generate associations between breastfeeding and medicines exposure that may not appear in the wider (non-volunteer or unselected) population (1). Accordingly, the cohorts and most databases identified in this scoping review are not suitable for estimating the prevalence of infant ADRs arising from breastfeeding: rather, they alert professionals and families to potential problems to be monitored. Absence of hospital prescribing data may have caused exposure misclassification, and studies focusing on people contacting information services include only healthy survivors (immortal time bias) [91].

Infant follow-up ranged from 2 weeks to 3 years. No education outcomes were reported. Of the large studies, only the MoBa study systematically reported long-term neurodevelopmental outcomes. Twelve studies reported various and disparate developmental outcomes and assessments.

Strengths and limitations of the review

This systematic review used a ‘wide-net’ approach to locate the primary surveillance data and identify the range of safety endpoints for a defined population, rather than focussing on a single safety endpoint [92] or information on each medicine category [67]. However, the terms “product surveillance” and “drug” identified articles on pesticides and recreational drugs. Study eligibility criteria, identification and selection of studies, data collection, appraisal, and synthesis were debated by all authors [92, 93], with due consideration for the differences between scoping reviews aiming to identify sources of data and systematic reviews aiming to answer clinical questions (S1 Checklist). The focus of the review on identifying databases with data on medicines AND infant outcomes AND breastfeeding led to omission of studies assessing only breastfeeding rates following medicines exposure [71, 72, 88], and some that did not use databases [37, 94]. Of the five European databases known to us as holding breastfeeding data, only the French databases appeared in our search [1].

Although included in the initial search strategy, papers in languages other than English (without English abstracts) were excluded, due to absence of tables of empirical data, and practical difficulties. We were unable to obtain some early papers, but none appeared to contain empirical data. We excluded studies on transfer of medicines through breastmilk where there was no information on infant outcomes.

In this scoping review, as anticipated, data on breastfeeding did not lend themselves to meta-analysis: outcomes, reporting methods, and study designs were heterogeneous. This complicated selection of a ‘risk of bias’ assessment instrument; however, since all studies related to exposure, we selected ROBINS-E, which is designed for studies of exposure [24, 95]. However, measurement of direction of bias was impractical [95]. Risk of bias was reported to illustrate the heterogeneity and paucity of the evidence rather than to influence the summary of the data [20]. Accordingly, our data summary, by failing to offer reassurance, serves to signpost the need for further research.


1. Breastfeeding dyads

Professionals caring for breastfeeding patients receiving prescription medicines should monitor breastfed infants for signs of known ADRs: for example, those using mental health medicines should observe for sleepiness, drowsiness and sedation, and, where necessary, venous blood samples should be arranged [61, 96]. Monitoring ranges from checking the infant’s mouth for oral thrush to weight and sleep charting to venous blood sampling. Although risks are unquantified, due to lack of data, developmental and physical monitoring needs to be more intensive than the standard 72 hours and 6 week physical assessments [97]. Excessive sleeping will impair optimal development and may lead to failure to gain weight and thrive. These signs and symptoms are subtle, and may be overlooked if not specifically monitored. A quiet infant that cries little and sleeps a lot may be viewed as easy to manage, particularly where cultural norms suggest: ‘a sleeping baby is a good baby’.

Breastfeeding patients prescribed mental health medicines or antiepileptics are at increased risk of ‘not breastfeeding’ and early discontinuation. Professionals should be aware of this risk, and advise and support accordingly. Whilst failure to initiate breastfeeding may be related to choice (often driven by worry and uncertainty regarding transfer of medicines), early discontinuations is unlikely to be attributable to confounding by indication, and should be recognised as a possible biological effect of prescribed medicines on milk supply.

2. Paucity of data

This review has identified few ongoing databases with breastfeeding data, and none reporting prescribing in hospitals. Any infant harms due to exposure via breastmilk are likely to be subtle, rendering the absence of long-term follow-up and educational outcomes critical. Families and professionals rely on established databases, such as LactMed, for information [67]; however, despite thorough searches, the databases often have little information to offer, and report only small case series. No sections of the population should be excluded from the protection afforded by timely collection and analysis of data on the safety of medicinal products [78]. However, the omission of breastfeeding data from most population databases indicates that there are few data to inform breastfeeding patients and those intending to breastfeed a) whether lactation will be affected by prescription medicines, and b) how medicines will affect breastfed infants. To return investment in population healthcare databases pharmacoepidemiologists should have good quality data to explore any relationships between medicines exposures, breastfeeding and short- and long-term infant outcomes.

Supporting information

S1 File. Registry of systematic reviews.

Review no.994.


S1 Table. Cohort studies in chronological order.



We should like to acknowledge the work of Naomi Marfell, Cardiff University School of Medicine, Cardiff, Wales, UK, and Stephen Storey, Librarian, Swansea University, Swansea, Wales UK in conducting searches, and Maike Tauchert, BBMRI-ERIC, Graz, Austria for reviewing.


Jordan S., Komninou S., Marfell N. 2020 Review of data sources for breastfeeding and medicine exposure. Research Registry. REGISTRY OF SYSTEMATIC REVIEWS/META-ANALYSES. Review registry number 994 (11.9.20).


  1. 1. Jordan S, Bromley R, D-M C, Given J; Komninou S; Loane M; et al. Breastfeeding, pregnancy, medicines, neurodevelopment, and population databases: the information desert. International Breastfeeding Journal. 2022;17(1):55. pmid:35915474.
  2. 2. Bartick M, Schwarz E, Green B, Jegier B, Reinhold A, Colaizy T, et al. Suboptimal breastfeeding in the United States: Maternal and pediatric health outcomes and costs. Matern Child Nutr. 2017;13(1):e12366. pmid:27647492
  3. 3. Güngör D, Nadaud P, LaPergola C. Infant milk-feeding practices and food allergies, allergic rhinitis, atopic dermatitis, and asthma throughout the life span: a systematic review [published correction appears in Am J Clin Nutr.;110(4):1041]. Am J Clin Nutr. 2019;109(Suppl_:772S-99S. doi: 10.1093/ajcn/nqy283.
  4. 4. Verstegen R, Anderson P, Ito S. Infant drug exposure via breast milk Br J Clin Pharmacol. 2020. p. 1–17.
  5. 5. Martin A, Bland RM, Connelly A, Reilly JJ. Impact of adherence to WHO infant feeding recommendations on later risk of obesity and non‐communicable diseases: systematic review. Maternal & Child Nutrition. 2016;12(3):418–27. pmid:26259927
  6. 6. Fraser A, Macdonald-Wallis C, Tilling K, Boyd A, Golding J, Davey Smith G, et al. Cohort Profile: the Avon Longitudinal Study of Parents and Children: ALSPAC mothers cohort. Int J Epidemiol. 2013;42:97–110. pmid:22507742
  7. 7. Olsen J, Melbye M, Olsen SF, Sørensen TI, Aaby P, Nybo Andersen A-M, et al. The Danish National Birth Cohort-its background, structure and aim. Scandinavian journal of public health. 2001;29(4):300–7. pmid:11775787
  8. 8. Lund-Blix N, Dydensborg Sander S, Størdal K. Infant Feeding and Risk of Type 1 Diabetes in Two Large Scandinavian Birth Cohorts. Diabetes Care. 2017;40:920–7. pmid:28487451
  9. 9. Jordan S, Watkins A, Storey M, Allen SJ, Brooks CJ, Garaiova I, et al. Volunteer Bias in Recruitment, Retention, and Blood Sample Donation in a Randomised Controlled Trial Involving Mothers and Their Children at Six Months and Two Years: A Longitudinal Analysis. PLoS ONE 2013;8:e67912. pmid:23874465
  10. 10. Cole S, Platt R, Schisterman E, Chu H, Westreich D, Richardson D, et al. Illustrating bias due to conditioning on a collider. International Journal of Epidemiology,. 2010;39:417–20. pmid:19926667
  11. 11. Yan R, Liu T, Peng Y, Peng X. Can statistical adjustment guided by causal inference improve the accuracy of effect estimation? A simulation and empirical research based on meta-analyses of case-control studies. BMC Med Inform Decis Mak. 2020;20:333. pmid:33308213
  12. 12. Griffith G, Morris T, Tudball M. Collider bias undermines our understanding of COVID-19 disease risk and severity. Nat Commun. 2020;11:5749. pmid:33184277
  13. 13. Kincaid H. Causal modeling, mechanism, and probability in epidemiology. In: McKay Illari P, Russo F, Williamson J, editors. Causality in the Sciences. Oxford: Oxford University Press; 2011. p. 170–90.
  14. 14. Elwert F, Winship C. Endogenous Selection Bias: The Problem of Conditioning on a Collider Variable. Annu Rev Sociol. 2014;40:31–53. pmid:30111904
  15. 15. Charlton R, Bettoli V, Bos HJ, Engeland A, Garne E, Gini R, et al. The limitations of some European healthcare databases for monitoring the effectiveness of pregnancy prevention programmes as risk minimisation measures. European journal of clinical pharmacology. 2018;74(4):513–20. pmid:29230493
  16. 16. Dictionary OE (Oxford English). Database UK:; 2022. accessed 21.3.23
  17. 17. Donaldson MS LK. Health Data in the Information Age: Use, Disclosure, and Privacy. Networks IoMUCoRHD, editor, 1994.
  18. 18. Altman D. Practical Statistics for Medical Research. London: Chapman & Hall; 1991.
  19. 19. Munn Z, Peters MD, Stern C, Tufanaru C, McArthur A, Aromataris E. Systematic review or scoping review? Guidance for authors when choosing between a systematic or scoping review approach. BMC medical research methodology. 2018;18(1):1–7.
  20. 20. Arksey H O’Malley L. Scoping studies: towards a methodological framework. International journal of social research methodology. 2005;8(1):19–32.
  21. 21. Jordan S, Komninou S, Marfell, N. Review of data sources for breastfeeding and medicine exposure.: Research Registry. REGISTRY OF SYSTEMATIC REVIEWS/META-ANALYSES.; 2020. Available from: accessed 21.3.23
  22. 22. Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. International Journal of Surgery. 2021;88:105906. pmid:33789826
  23. 23. Tricco AC, Lillie E, Zarin W, O’Brien KK, Colquhoun H, Levac D, et al. PRISMA extension for scoping reviews (PRISMA-ScR): checklist and explanation. Annals of internal medicine. 2018;169(7):467–73. pmid:30178033
  24. 24. Group R-ED. Risk Of Bias In Non-randomized Studies—of Exposure (ROBINS-E). 2022. Available from: Available from: accessed 21.3.23
  25. 25. Rogan WJ, Gladen BC, McKinney JD, Carreras N, Hardy P, Thullen J, et al. Polychlorinated biphenyls (PCBs) and dichlorodiphenyl dichloroethene (DDE) in human milk: Effects on growth, morbidity, and duration of lactation. American Journal of Public Health. 1987;77:1294–7. pmid:3115123.
  26. 26. Schmidt K, Olesen OV, Jensen PN. Citalopram and breast-feeding: Serum concentration and side effects in the infant. Biological Psychiatry. 2000;47:164–5. pmid:10664835.
  27. 27. Gardiner SJ, Kristensen JH, Begg EJ, Hackett LP, Wilson DA, Ilett KF, et al. Transfer of olanzapine into breast milk, calculation of infant drug dose, and effect on breast-fed infants. American Journal of Psychiatry: American Psychiatric Publishing; 2003; 160(8):1428–31. pmid:12900304
  28. 28. Veiby G, Engelsen B, Gilhus N. Early Child Development and Exposure to Antiepileptic Drugs Prenatally and Through Breastfeeding: A Prospective Cohort Study on Children of Women With Epilepsy. JAMA Neurol. 2013;70:1367–74. pmid:24061295
  29. 29. Ko JY, Tong VT, Bombard JM, Hayes DK, Davy J, Perham-Hester KA. Marijuana use during and after pregnancy and association of prenatal use on birth outcomes: A population-based study. Drug and Alcohol Dependence. 2018;187:72–8. pmid:29627409.
  30. 30. Soussan C, Gouraud A, Portolan G, Jean-Pastor MJ, Pecriaux C, Montastruc JL, et al. Drug-induced adverse reactions via breastfeeding: A descriptive study in the french pharmacovigilance database. European Journal of Clinical Pharmacology. 2014;70:1361–6. pmid:25183382.
  31. 31. Goldstein DJ, Corbin LA, Fung MC. Olanzapine-exposed pregnancies and lactation: Early experience. Journal of Clinical Psychopharmacology. 2000;20:399–403. pmid:10917399.
  32. 32. Noseda R, Bedussi F, Gobbi C, Zecca C, Ceschi A. Safety profile of erenumab, galcanezumab and fremanezumab in pregnancy and lactation: Analysis of the WHO pharmacovigilance database. Cephalalgia. 2021;41(7):789–98. pmid:33435709
  33. 33. Kaplan S, Zeygarnik M, Stern T. Pregnancy, Fetal, and Infant Outcomes Following Maternal Exposure to Glatiramer Acetate During Pregnancy and Breastfeeding. Drug Safety. 2022;45(4):345–57. pmid:35297004
  34. 34. Brunner E, Falk DM, Jones M, Dey DK, Shatapathy CC. Olanzapine in pregnancy and breastfeeding: A review of data from global safety surveillance. BMC Pharmacology and Toxicology: BMC Pharmacol Toxicol; 2013;14:38. pmid:23902726
  35. 35. Kronenfeld N, Baran Tz, Berlin M, Karra N, Dinavitser N, Cohen R, et al. Chronic use of psychotropic medications in breastfeeding women: Is it safe? PLoS ONE. 2018;13:e0197196. pmid:29782546.
  36. 36. Gilad O, Merlob P, Stahl B, Klinger G. Outcome of infants exposed to olanzapine during breastfeeding. Breastfeeding Medicine. 2011;6:55–8. pmid:21034242.
  37. 37. Gorman JR, Kao K, Chambers CD. Breastfeeding among women exposed to antidepressants during pregnancy. Journal of Human Lactation. 2012;28:181–8. pmid:22344850.
  38. 38. Ito S, Blajchman A, Stephenson M, Eliopoulos C, Koren G. Prospective follow-up of adverse reactions in breast-fed infants exposed to maternal medication. American Journal of Obstetrics and Gynecology. 1993;168:1393–9. pmid:8498418.
  39. 39. Crume TL, Juhl AL, Brooks-Russell A, Hall KE, Wymore E, Borgelt LM. Cannabis Use During the Perinatal Period in a State With Legalized Recreational and Medical Marijuana: The Association Between Maternal Characteristics, Breastfeeding Patterns, and Neonatal Outcomes. Journal of Pediatrics. 2018;197:90–6. pmid:29605394.
  40. 40. Dryden-Peterson S, Shapiro RL, Hughes MD, Powis K, Ogwu A, Moffat C, et al. Increased risk of severe infant anemia after exposure to maternal HAART, Botswana. Journal of Acquired Immune Deficiency Syndromes. 2011;56:428–36. pmid:21266910.
  41. 41. Chaves RG, Lamounier JA, César CC. Automedicação em nutrizes e sua influência sobre a duração do aleitamento materno. Jornal de Pediatria. 2009;85:129–34.
  42. 42. Froescher W, Eichelbaum M, Niesen M, Dietrich K, Rausch P. Carbamazepine levels in breast milk. Therapeutic Drug Monitoring. 1984;6:266–71. pmid:6390794.
  43. 43. Lee A, Woo J, Ito S. Frequency of infant adverse events that are associated with citalopram use during breast-feeding. American Journal of Obstetrics and Gynecology. 2004;190:218–21. pmid:14749663.
  44. 44. Kelly LE, Poon S, Madadi P, Koren G. Neonatal benzodiazepines exposure during breastfeeding. Journal of Pediatrics. 2012;161:448–51. pmid:22504099.
  45. 45. Chinnatamby S. Effects of ’nordiol’ on fertility and lactation: Some preliminary observations. Current Medical Research and Opinion. 1973;1:376–8. pmid:4272507.
  46. 46. Boutroy MJ, Bianchetti G, Dubruc C, Vert P, Morselli PL. To nurse when receiving acebutolol: Is it dangerous for the neonate? European Journal of Clinical Pharmacology. 1986;30:737–9. pmid:3770068.
  47. 47. Lam J, Kelly L, Matok I, Ross CJD, Carleton BC, Hayden MR, et al. Putative association of ABCB1 2677G>T/A with oxycodone-induced central nervous system depression in breastfeeding mothers. Therapeutic Drug Monitoring. 2013;35:466–72. pmid:23783165.
  48. 48. Sinha SK, Kishore MT, Thippeswamy H, Kommu JVS, Chandra PS. Adverse effects and short-term developmental outcomes of infants exposed to atypical antipsychotics during breastfeeding. Indian Journal of Psychiatry. 2021;63(1):52. pmid:34083820
  49. 49. Merlob P, Stahl B, Sulkes J. Paroxetine during breast-feeding: Infant weight gain and maternal adherence to counsel. European Journal of Pediatrics. 2004;163:135–9. pmid:14745552.
  50. 50. Heikkinen T, Ekblad U, Kero P, Ekblad S, Laine K. Citalopram in pregnancy and lactation. Clinical Pharmacology and Therapeutics. 2002;72:184–91. pmid:12189365.
  51. 51. Yoshida K, Smith B, Craggs M, Channi Kumar R. Fluoxetine in breast-milk and developmental outcome of breast-fed infants. British Journal of Psychiatry. 1998;172:175–9. pmid:9519072.
  52. 52. Ito S, Moretti M, Lian M, Koren G. Initiation and duration of breast-feeding in women receiving antiepileptics. American Journal of Obstetrics and Gynecology. 1995;172:881–6. pmid:7892879.
  53. 53. Passmore C, McElnay J, Rainey E, D’Arcy P. Metronidazole excretion in human milk and its effect on the suckling neonate. British Journal of Clinical Pharmacology. 1988;26:45–51. pmid:3203060.
  54. 54. Yoshida K, Smith B, Craggs M, Kumar RC. Investigation of pharmacokinetics and of possible adverse effects in infants exposed to tricyclic antidepressants in breast-milk. Journal of Affective Disorders. 1997;43:225–37. pmid:9186793.
  55. 55. Moroni S, Marson ME, Moscatelli G, Mastrantonio G, Bisio M, Gonzalez N, et al. Negligible exposure to nifurtimox through breast milk during maternal treatment for Chagas Disease. PLoS Neglected Tropical Diseases. 2019;13:e0007647. pmid:31415566.
  56. 56. Viguera AC, Newport DJ, Ritchie J, Stowe Z, Whitfield T, Mogielnicki J, et al. Lithium in breast milk and nursing infants: Clinical implications. American Journal of Psychiatry. 2007;164:342–5. pmid:17267800.
  57. 57. Ilett KF, Hackett LP, Kristensen JH, Rampono J, Kohan R. Estimation Of Infant Dose And Assessment Of Breastfeeding Safety For Escitalopram Use In Postnatal Depression. Therapeutic Drug Monitoring. 2005;27:248.
  58. 58. Brown R, Dakkak H, Seabrook J. Is breast best? Examining the effects of alcohol and cannabis use during lactation. Journal of neonatal-perinatal medicine. 2018;11(4):345–56. pmid:29843260
  59. 59. Matheson I, Samseth M, Loberg R, Faegri A, Prentice A. Milk transfer of phenoxymethylpenicillin during puerperal mastitis. British journal of clinical pharmacology. 1988;25(1):33–40. pmid:3130891
  60. 60. Anderson PO. Drugs in lactation. Pharmaceutical research. 2018;35(3):1–13. pmid:29411152
  61. 61. Gehrmann A, Fiedler K, Leutritz AL, Koreny C, Kittel-Schneider S. Lithium medication in pregnancy and breastfeeding—a case series. Medicina. 2021;57(6):634. pmid:34207460
  62. 62. Bovbjerg ML, Uphoff AE, Rosenberg KD. Two-Year Test-Retest Reliability of the Breastfeeding Duration Question Used By the Pregnancy Risk Assessment Monitoring System (PRAMS): Implications for Research. Maternal and Child Health Journal. 2021;25(7):1126–35. pmid:33909204
  63. 63. Ahluwalia IB, Helms K, Morrow B. Assessing the validity and reliability of three indicators self-reported on the pregnancy risk assessment monitoring system survey. Public Health Reports. 2013;128(6):527–36. pmid:24179264
  64. 64. Li R, Scanlon KS, Serdula MK. The validity and reliability of maternal recall of breastfeeding practice. Nutrition reviews. 2005;63(4):103–10. pmid:15869124
  65. 65. Roque Pereira L, Durán CE, Layton D, Poulentzas G, Lalagkas P-N, Kontogiorgis C, et al. A Landscape Analysis of Post-Marketing Studies Registered in the EU PAS Register and ClinicalTrials. gov Focusing on Pregnancy Outcomes or Breastfeeding Effects: A Contribution from the ConcePTION Project. Drug safety. 2022;45(4):333–44. pmid:35357659
  66. 66. Lawrence R, Schaefer C. General commentary on drug therapy and drug risk during lactation. In: Schefer C, editor. Drugs during pregnancy and laction. Amsterdam: Elsevier; 2015. p. 637–48.
  67. 67. Drugs and Lactation Database [Internet]. National Library of Medicine (US). 2006. Available from: accessed 21.3.23
  68. 68. BNF JFC. BNF 83: March 2022 -September 2022. London: Pharmaceutical Press; 2022.
  69. 69. Meador KJ, Baker GA, Browning N, Cohen MJ, Bromley RL, Clayton-Smith J, et al. Breastfeeding in children of women taking antiepileptic drugs: cognitive outcomes at age 6 years. JAMA pediatrics. 2014;168(8):729–36. pmid:24934501
  70. 70. Shawahna R, Zaid L. Concentrations of antiseizure medications in breast milk of lactating women with epilepsy: A systematic review with qualitative synthesis. Seizure. 2022;98:57–70. Epub 2022/04/16. pmid:35427849.
  71. 71. Jordan S, Davies G, Thayer D, Tucker D, Humphreys I. Antidepressant prescriptions, discontinuation, depression and perinatal outcomes, including breastfeeding: a population cohort analysis. Plos One. 2019; 15(12):e0242489. pmid:31738813
  72. 72. Grzeskowiak LE, Saha MR, Nordeng H, Ystrom E, Amir LH. Perinatal antidepressant use and breastfeeding outcomes: Findings from the Norwegian Mother, Father and Child Cohort Study. Acta Obstetricia et Gynecologica Scandinavica. 2022; 101(3):344–354. pmid:35170756
  73. 73. Saha M, Ryan K, Amir L. Postpartum women’s use of medicines and breastfeeding practices: a systematic review. Int Breastfeed J. 2015;10:28. pmid:26516340
  74. 74. Spiesser-Robelet L, Brunie V, de Andrade V, Gagnayre R. Knowledge, representations, attitudes, and behaviors of women faced with taking medications while breastfeeding: A scoping review. Journal of Human Lactation. 2017;33(1):98–114.
  75. 75. Wang J, Johnson T, Sahin L, Tassinari M, Anderson P, Baker T, et al. Evaluation of the safety of drugs and biological products used during lactation: workshop summary. Clinical Pharmacology & Therapeutics. 2017;101(6):736–44. pmid:28510297
  76. 76. FDA. Pregnancy, Lactation, and Reproductive Potential: Labeling for Human Prescription Drug and Biological Products—Content and Format Guidance for Industry 2020. Available from: Accessed 21.3.23
  77. 77. European Medicines Agency. Guideline on risk assesment of medicinal products on human reproduction and lactation: From data to labelling 2008. Available from: Accessed 21.3.23
  78. 78. European Medicines Agency. Guideline on good pharmacovigilance practices (GVP) 3 Product- or Population-Specific Considerations III: Pregnant and 4 breastfeeding women. EMA/653036/2019 DRAFT FOR PUBLIC CONSULTATION EMA, Heads of Medicines agencies 4.12.19, Amsterdam Accessed 2.11. 2019.
  79. 79. Yalcin N, Flint RB, van Schaik RHN, Simons SHP, Allegaert K. The Impact of Pharmacogenetics on Pharmacokinetics and Pharmacodynamics in Neonates and Infants: A Systematic Review. Pharmgenomics Pers Med. 2022;15:675–96. Epub 2022/07/08. pmid:35795337; PubMed Central PMCID: PMC9252316.
  80. 80. Hill AB. The environment and disease: association or causation?: Sage Publications; Proc R Soc Med. 1965 May;58(5):295–300. PMID: pmid:14283879; PMCID: PMC1898525.
  81. 81. Gallini F, Maggio L, Romagnoli C, Marrocco G, Tortorolo G. Progression of renal function in preterm neonates with gestational age≤ 32 weeks. Pediatric nephrology. 2000;15(1):119–24.
  82. 82. Rodieux F, Wilbaux M, van den Anker JN, Pfister M. Effect of kidney function on drug kinetics and dosing in neonates, infants, and children. Clinical pharmacokinetics. 2015;54(12):1183–204. pmid:26138291
  83. 83. Rawson NS D’Arcy C. Healthcare databases for drug safety research: data validity assessment remains crucial. Drug Safety. 2018;41(9):829–33.
  84. 84. Morales DR, Slattery J, Evans S, Kurz X. Antidepressant use during pregnancy and risk of autism spectrum disorder and attention deficit hyperactivity disorder: systematic review of observational studies and methodological considerations. BMC Med. 2018; 16(1):6. pmid:29332605; PMCID: PMC5767968.
  85. 85. Chambers CD. Value of the small cohort study including a physical examination for minor structural defects in identifying new human teratogens. Congenital Anomalies. 2011;51(1):16–20. pmid:21158951
  86. 86. Hazell L, Shakir SA. Under-reporting of adverse drug reactions. Drug safety. 2006;29(5):385–96.
  87. 87. Hakobyan L, Haaijer-Ruskamp FM, de Zeeuw D, Dobre D, Denig P. A review of methods used in assessing non-serious adverse drug events in observational studies among type 2 diabetes mellitus patients. Health and quality of life outcomes. 2011;9(1):1–9. pmid:21958008
  88. 88. Davies G, Jordan S, Thayer D, Tucker D, Humphreys I. Medicines prescribed for asthma, discontinuation and perinatal outcomes, including breastfeeding: A population cohort analysis. PLOS ONE 2020;15. pmid:33296383
  89. 89. Draper NR, Smith H. Applied regression analysis: John Wiley & Sons; 1998.
  90. 90. Sendor R, Stürmer T. Core Concepts in Pharmacoepidemiology: Confounding by Indication and the Role of Active Comparators. Pharmacoepidemiology and drug safety. 2022; 31(3):261–269. pmid:35019190
  91. 91. ENCePP; The European Network of Centres for Pharmacoepidemiology and Pharmacovigilance. Guide on Methodological Standards in Pharmacoepidemiology (Revision 10). EMA/95098/2010 2022. Available from: Accessed 21.3.23
  92. 92. ENCePP; The European Network of Centres for Pharmacoepidemiology and Pharmacovigilance. Annex 1 to the Guide on Methodological Standards in Pharmacoepidemiology 2022. Available from: Accessed 21.3.23
  93. 93. Whiting P, Savović J, Higgins JP, Caldwell DM, Reeves BC, Shea B, et al. ROBIS: a new tool to assess risk of bias in systematic reviews was developed. Journal of clinical epidemiology. 2016;69:225–34. pmid:26092286
  94. 94. Venkatesh K, Castro V, Perlis R. Impact of antidepressant treatment during pregnancy on obstetric outcomes among women previously treated for depression: An observational cohort study. J Perinatol. 2017;37:1003–9. pmid:28682318
  95. 95. Bero L, Chartres N, Diong J, Fabbri A, Ghersi D, Lam J, et al. The risk of bias in observational studies of exposures (ROBINS-E) tool: concerns arising from application to observational studies of exposures. Syst Rev. 2018;7(1):242. Epub 2018/12/24. pmid:30577874; PubMed Central PMCID: PMC6302384.
  96. 96. Dathe K, Schaefer C. The use of medication in pregnancy. Deutsches Ärzteblatt International. 2019;116(46):783. pmid:31920194
  97. 97. Welsh Goverment. An Overview of the Healthy Child Wales Programme. 2016. Available from: accessed 21.3.23