Families need support to reduce the risk of adverse drug reactions from breastfeeding

Families and healthcare professionals often ask pharmacists if a medicine is compatible with breastfeeding; however, despite considerable investment in pharmacovigilance and the capture of healthcare data in population databases, often the only information on medicines and breastfeeding emanates from small case series or pharmacokinetic studies, and is therefore a suboptimal evidence-base for advice​[1]​.  

Most maternal medicines do not adversely affect most breastfed infants. Some — such as paracetamol, when prescribed alone — are ‘generally regarded as safe’, while others are associated with self-limiting adverse effects, typically diarrhoea or oral thrush following antibiotic exposure​[2,3]​. However, adverse drug reactions (ADRs) of varying severity from breastmilk exposure are widely reported, some of which are life-threatening, including apnoea and meningeal haemorrhage​[4–7]​. Long-term sequelae are rarely reported. 

Susceptibility to adverse drug reactions

Currently, there are insufficient data to predict which infants will be harmed, in the short or long term. The vulnerability of an infant depends on maturity, health status, genetics, timing of breastfeeding and co-exposures. 

Most breastfed infants’ ADRs mirror adult ADRs and are dose-related, emanating from the combined effects of low clearance and high transfer into breastmilk.

Low clearance

The immature renal and hepatic function of preterm or sick infants leaves them vulnerable to drug accumulation, even when healthy term infants are able to clear the medicine effectively, but little information is available. Some enzymes do not function during the first months of life, such as those metabolising caffeine​[6,8]​. From 30 to 34 weeks’ gestational age, other enzymes, such as those metabolising morphine, display the same polymorphisms and activity range as adults​[9,10]​. Similarly, variations in codeine and oxycodone metabolism may explain why only ~20% infants experience drowsiness following breastmilk exposure​[11,12]​.    

High transfer

Transfer from maternal plasma into breastmilk and infants’ circulation is highest in the first week postpartum and variable. For example, the milk–plasma ratio for venlafaxine ranges 0.85–4.85 (mean 2.59) and the lamotrigine and zonisamide infant–maternal blood concentrations range 0.6–90.3% (median 28.9%) and 35.2–125.3% (median 44.2%), respectively​[13,14]​.  

Some 25–30% of ADRs in breastfed infants are attributable to unpredictable hypersusceptibility or hypersensitivity responses​[6]​. For example, neutropoenia has been reported in association with carbimazole, apnoea with opioids and agranulocytosis in one of four infants exposed to clozapine​[7,15]​. 

No long-term data

Variations in concentration of antidepressants appeared greater in hind milk than fore milk samples, possibly owing to the higher lipid content of hind milk​[13]​. Case reports of infant harm are often associated with polypharmacy: for example, an infant exposed to olanzapine, clonazepam, droperidol, sertraline, thioridazine and valproic acid had impaired intellectual development​[16]​.  

In the absence of data from randomised controlled trials (RCTs) and whole-population prescribing databases, current practice recommendations rest on case series that suggest the potential for (at least) short-term ADRs in breastfed infants from: opioids, clozapine, amisulpride, combinations of central nervous system depressants, amiodarone, oral retinoids, radioiodine, topical and systemic free iodine, chemotherapy, all sedating medicines (including some antihistamines), quinolones, aspirin, gold salts, lithium, ergot derivatives, barbiturate derivatives, recreational drugs and alcohol in most doses​[5,6,17–19]​. 

However, variations in susceptibility indicate that quantifying the risk from rare but serious adverse events and identifying vulnerable subgroups will necessitate large studies to account for moderators, such as age at exposure, genotypes, comorbidities and time-varying confounders, such as co-prescribing and co-exposures​[1]​. Without reassuring long-term data, anxieties over short-term or rare ADRs may deprive infants of the benefits of breastfeeding. 

Developmental outcomes

Breastfeeding benefited children’s development in an RCT and observational studies​[20,21]​. Short-term exposure to psychotropic medicines or opioids via breastmilk may cause sedation, irritability, restlessness, diarrhoea and suboptimal weight gain, but how these relate to childhood development is unknown​[5,22]​. 

Long-term data are scarce, with only five population databases in Europe holding data on breastfeeding plus medicines plus developmental outcomes — these will be explored in the ConcePTION study​[1]​. Recruited cohorts of infants exposed to antiepileptics in utero plus via breastmilk report that the additional postnatal exposure from breastfeeding is not disadvantageous; however, data regarding phenytoin (n=36) are less reassuring​[23,24]​. 

Resolving the dilemma “should those prescribed long-term medicines breastfeed?” will require long-term research in whole populations. If the whole population is not studied, findings may be distorted by collider and volunteer selection bias and relate only to volunteer samples, which usually represent the more affluent sections of the population​[1,25]​.  

Successful breastfeeding

The complex physiology of lactation is also vulnerable to disruption by medicines that affect serotonergic pathways (including antidepressants); antagonise prolactin (amphetamines, oestrogens, ergot derivatives, aripiprazole, nicotine, antihistamines [promethazine, diphenhydramine, chlorpheniramine], injected corticosteroids, possibly diuretics, progesterone, tamoxifen); or reduce oxytocin release (alcohol, opioids, sympathomimetics [pseudoephedrine, phenylephrine, epinephrine], anticholinergics, antihistamines, antidepressants)​[4,26–28]​. Exposure to some medicines in late pregnancy, labour and postpartum affects breastfeeding rates: a dose–response effect is postulated​[28–31]​.  

Spontaneous reporting is important in signal generation but may only capture 5–10% of ADRs​[32]​. Women may simply discontinue breastfeeding without reporting any suspicion of a link to a prescription medicine. Therefore, identifying associations between medicines and breastfeeding rates will need population-based approaches. 

Data blind spot 

Current data on medicines and breastfeeding are inconclusive. When marketed, most medicines have no data on breastfeeding: this is not obligatory​[33]​. Consequently, for some medicines (including ciclosporin, mycophenolate mofetil, belatacept, fingolimod, and other immunosuppressants), formularies indicate “manufacturer advises avoid”, sometimes based on “no information available”​[34]​. 

Few medicines are licensed for use during lactation and there may be no reports of safe administration to neonates​[17]​. The Cumberlege Report details how case reports of valproate teratogenicity from the 1980s were not actioned for decades — analyses of large population databases would militate against such oversights, but first, whole-population data on breastfeeding must be collected​[1,35]​. 

Support and monitoring

What should be done while we await more evidence? There are several approaches that could be beneficial in this area.

First, women prescribed long-term medicines are less likely to breastfeed and so the uncertainty and concern surrounding lack of information on infant ADRs, coupled with physiological difficulties, indicate that support may be needed to optimise breastfeeding​[3,36]​. Existing analyses offer sufficient evidence for using prescription records to trigger low-risk interventions, including additional breastfeeding support to the 15% of women prescribed antidepressants in the infant’s first year​[29,30,37,38]​.  

Second, people receiving long-term repeat prescriptions should be regularly monitored for their medicine’s adverse effects and effectiveness​[30]​. This structured, formalised profiling approach should be extended to breastfed infants. Routine surveillance by midwives and health visitors is focused on immunisations, public health, emotional development and parenting concerns: the physical assessments at 72 hours and 6 weeks are not designed to monitor ADRs and are insufficient for noting sedation, irritability or other problems that may have developed in association with medicine administration​[39]​. 

Regular medicine-specific monitoring would detect problems early enough to pre-empt serious problems and offer parents reassurance regarding the safety of medicines during breastfeeding. This approach, with laboratory monitoring and genetic testing as indicated, is advocated by specialist teams caring for women prescribed lithium. Until more data are available, monitoring could be extended to other medicines known to affect breastfed infants, particularly opioids and psychotropics​[40,41]​. This would offer a response to the dilemma “do I take the medicine for my own needs or breastfeed my baby?”.