To inform or not to inform
When valproate was introduced to the European market in the 1970s the regulatory authorities were already aware that the drug was teratogenic in animals, and scientific reports soon emerged describing spina bifida and developmental delay in children born to women who had used the drug (Figure 1). Fetal valproate syndrome was first described in 1984 (9). The risk was communicated to prescribing doctors, but not directly to patients via patient information leaflets. There was also no recommendation from the regulatory authorities to change treatment prior to pregnancy. Advisory experts considered that informing patients directly could cause fear and reduce adherence, and that it was preferable for the information to be conveyed by a doctor (9).
When valproate was introduced to the European market in the 1970s the regulatory authorities were already aware that the drug was teratogenic in animals, and scientific reports soon emerged describing spina bifida and developmental delay in children born to women who had used the drug
There was also no requirement for additional safety data. Even after animal studies in the 1990s demonstrated autism-like behaviour following in utero exposure to valproate, regulators failed to call for systematic studies of risk in pregnancy (9). The finding that pregnant mice exposed to valproate produced offspring with autistic-like traits gave rise to the 'valproate animal model of autism', which was quickly adopted in behavioural research. However, this new insight did not lead to a greater focus on reproductive safety in humans.
Doctors were expected to explain the risks associated with valproate to patients, but in practice many women were not informed and used the drug at high doses during pregnancy (9). Before 2004, there was no mention in patient information leaflets that valproate was not recommended during pregnancy (9). The risks and the requirement to inform women were only clearly communicated by regulatory authorities in the period 2012–15. This followed studies showing that children exposed in utero had an approximately 9–25 % risk of congenital malformations, a two- to five-fold increased risk of autism spectrum disorders, and an IQ ten points lower on average than that of unexposed children (10). These risk communication measures had limited impact on patient awareness (11), and three years later the EMA introduced a comprehensive pregnancy prevention programme for valproate users.
Valproate should not be used in girls and women of reproductive age unless other treatments are ineffective or not tolerated. Users must be monitored by a specialist, use effective contraception, undergo pregnancy testing and complete an annual risk assessment in which healthcare personnel review specified information points with the patient.
In the years that followed, legal action was taken against manufacturers, authorities and individual doctors. In France, 1660 patients have received nearly €80 million in compensation. Pharmaceutical companies have been ordered to pay damages but argue that the teratogenic risks have been communicated to doctors since the 1980s, and that it is the regulatory authorities who determine what is included in patient information leaflets (12). In the United Kingdom, significant government compensation is being considered for up to 20,000 individuals harmed by valproate (13).