Effectiveness of risk minimisation measures for valproate: A cross‐sectional survey among physicians in Europe

Abstract Purpose This study evaluated the effectiveness of risk minimisation measures (RMMs) implemented following the 2014 referral for valproate in Europe. Methods Cross‐sectional survey was conducted over 2‐month period in 2016 among physicians who prescribed valproate in France, Germany, the United Kingdom, Spain and Sweden. The web‐based questionnaire included five endpoints to evaluate physicians' knowledge on (a) prescribing valproate only for epilepsy and bipolar disorder in women if other treatments were ineffective or not tolerated; (b) ensuring supervision by experienced physicians while treating these conditions; (c) considering alternative treatments for women planning pregnancy, regular review of treatment needs and re‐assessing the benefit–risk balance in women and girls reaching puberty; (d) informing patients about the risks of taking valproate during pregnancy and (e) advising women on effective contraception during their treatment. Results Among 1153 physicians, 95.5% responded prescribing valproate for epilepsy and bipolar disorder in women only if other treatments are ineffective/not tolerated; 66.5% supervised while treatment; 76.6% considered alternative treatments for women planning pregnancy; 92.1% informed patients about the risks of taking valproate during pregnancy and 94.4% advised patients on the use of effective contraception during its treatment. Overall, 25.8% physicians recalled receiving both educational material (EM) and Dear Healthcare Professional Communication (DHPC). All endpoint rates were higher for physicians who acknowledged receipt of both DHPC and EM compared to physicians who did not receive them. Conclusions Although results varied across geography and physician speciality, majority of physicians had good knowledge about the indication and safety aspects of prescribing and using valproate.


| INTRODUCTION
Valproate and related substances (ie, sodium valproate, valproic acid, magnesium valproate, valproate, valpromide and valproate semi sodium) were first licensed in Europe for the indication of epilepsy and bipolar disorder in the 1960s. Valproate is also indicated for prophylaxis of migraine attack in few European countries. However, following approval of valproate, concerns were raised about the risk of teratogenicity associated with valproate use during pregnancy. 1 A systematic review and meta-analysis in pregnant women receiving antiepileptics drugs (AEDs) for epilepsy showed valproate monotherapy to be associated with the highest incidence of congenital malformations (10.7%), followed by phenytoin (7.3%). 2 Also, children born to women who were exposed to valproate and/or related substance during pregnancy were at risk of serious developmental disorders related to language, intellectual and memory abilities. [2][3][4][5] Considering the evidence associated with high risk of congenital (e) in countries where medicines constituting valproate are authorised for the prevention of migraine, women must not use valproate for preventing migraine when they are pregnant. 8 Pregnancy should be excluded before starting treatment for migraine, and women should use effective contraception; (f) physicians who prescribe valproate should provide women with complete information enabling them to fully understand of the risks of the treatment regarding pregnancy. 8 As part of the risk management plan for valproate, a postauthorisation safety study (PASS) survey was conducted among healthcare professionals (HCPs), in parallel to a real-world drug utilisation study (DUS). The objective of this PASS survey was to evaluate the effectiveness of the RMMs and specifically to evaluate the knowledge and behaviour of HCP regarding the latest prescribing conditions and safety information about valproate and related substances. We aimed to recruit a total of 1387 physicians (324 each from France, Germany and the UK, 260 from Spain and 155 from Sweden), in order to assess at least 1067 analysable questionnaires, assuming a 30% rate of non-analysable questionnaires. The sample size was calculated to achieve a minimum of 3% precision level for a proportion of 50% within 95% CI. Invitations to complete the survey were sent via e-mail to a randomly selected sample of physicians. Physicians were identified according to their speciality from a comprehensive list in the IQVIA OneKey database, which is a global reference healthcare professional database providing comprehensive and integrated information of more than 15 million HCPs across 100 countries. OneKey continuously updates the accuracy and level of details in the database through a dedicated team of 700 experts. 9 The physician selection was random, and no pre-registration was required by the physicians. Furthermore, there was no involvement of any pharmaceutical company. Physicians accessed the survey in the study website through a distinct identification number and password. All physicians who participated in the webbased survey were requested to complete an informed consent form. • Physicians acknowledging receipt of RMMs materials had better knowledge regarding appropriate prescribing of valproate compared to physicians who did not acknowledge receipt.
• Overall, majority of physicians correctly responded to the five key primary endpoints ranging from 66.5% to 95.5%. the targeted physicians were called and sent e-mail reminders through experienced operators before they were considered not reachable. The questionnaire was provided in English and local language in each participating country. The survey was translated to local languages using forth and back method, that is, a person translated it from English to the destination language and then another person translated back the questionnaire to English. It was then compared for inconsistencies with the original English version. The questionnaire was pre-tested for its consistency, comprehensibility and appropriateness of medical terms among six physicians before its implementation.

| Survey methodology
To minimise the opportunity for physicians to search for answers or to change their replies upon the content in successive questions, physicians could not revise/change the answers to preceding questions. The time for its completion was up to 20 minutes.

| Statistical methods
All analyses were descriptive in nature. The primary analysis included the population set of all physicians who met the eligibility criteria and answered all questions in the questionnaire. No statistical comparison were made, since no hypothesis testing was performed in the current study. Categorical variables were described as the total number and relative percentage per category. Quantitative variables were described with standard statistics including mean and SD. Missing data were displayed when present (value > 0). A random stratified sampling method was applied within stratification by country and speciality. For generalisability, the survey results were weighted to project the proportion (or results) to the entire population within the country and by speciality. A weight variable was applied to the analysable units during the calculation of result, to correct any overor under-sampling that may have occurred for a country or speciality.
Analyses were performed overall, by speciality and by country. Data analyses were conducted using the SAS software V9.4 (SAS Institute, Inc., Cary, North Carolina). were from the UK. The overall response rate among physicians was highest in Spain (99.7%) and lowest in France (20.2%) ( Table 2).

| Demographics and clinical practice characteristics
Overall, majority of the surveyed physicians with complete and analysable questionnaire were in the 40 to 59-year age group (70.6%) and had more than 10 years of experience (88.5%). Overall, GPs or family physicians (34.3%) and hospital-based (46.0%) constituted the largest percentage among all specialities and settings. The mean (±SD) number of valproate prescriptions to female patients per physician prescribed within the previous year was 48.2 (±100.5) ( Table 3).
In the UK, the highest proportion of physicians were in the age group of 40 to 49 years. In all countries, except Spain, GPs were mainly office-based.  (Figure 1). Among specialists, neurologists were less knowledgeable of this condition (80.0%-87.5%) ( Figure S1).

| Knowledge of the prescribing conditions and safety information warnings of valproate
Among all specialities, 78.7% of the physicians weighed the benefits against risks while prescribing valproate and/or related substances when a woman planned her pregnancy. Across all specialities, the weighing of benefit against risk assessment when a woman planned her pregnancy was highest among psychiatrists (89.8%) and least among GPs (76%) (Figure 2A).  Figure 2B).     20,21 The potential for selection bias due to voluntary participation of targeted physicians is another inherent bias in webbased surveys which affects the quality of the survey and leads to unreliable results and inferences. 22,23 Hence, to quantify any selection bias, the distribution of stratification criterion of physicians was compared between participants and non-participants. Furthermore, in order to reach the target per physician speciality and country, efforts were made to contact the required specialists.

| Acknowledgement of receipt of DHPC and EM
This survey reflected physician's knowledge and perception about treatment rather than the actual prescribing pattern of valproate in WCBP. Therefore, a DUS was parallelly conducted to confirm the actual prescribing patterns among WCBP. The findings of both studies, other utilisation studies and clinical insights led to 2018 EMA requirements to conduct additional DUS and survey to characterise the nature and the extent of the risks by valproate and/or related substances, following the strengthening of RMMs. 24 In summary, the study revealed that despite geographical and speciality-based variations for the knowledge of indication and safety aspects of valproate, majority of the surveyed physicians appropriately understood the prescribing indications of valproate and related substances in relation to WCBP and pregnant women.

ETHICAL STATEMENT
The survey followed the regulatory and ethical requirements of each country. The survey complied with the module VIII of the good pharmacovigilance practices (GVP). The study followed the European Pharmaceutical Marketing Research Association (EphMRA) code of conduct guidelines for all countries. Due to the survey nature of this study, ethical review was not required. Additionally, no legal approvals or information was required for the five countries.