Search

Rationale and background In March 2018, the European risk minimisation measures (RMMs) with a Pregnancy Prevention Program (PPP) for valproate-containing medicines was updated. A pharmacoepidemiological study was conducted using longitudinal data collected in five electronic health care databases from four EU countries and the UK to investigate the use of valproates authorised in the EU before and after implementation of the 2018 revised measures for pregnancy prevention in clinical practice, and effectiveness of the 2018 intervention. Objectives Objective 1: To determine drug utilisation and prescription patterns of valproate-containing medicinal products in females of childbearing potential, and to investigate whether significant changes in prescribing patterns occurred (pre-/post-intervention). Objective 2: To determine prescribers’ compliance with the recommendations in the Summary of Products Characteristics (SmPC) for valproate-containing medicinal products, by indication, age group, duration of use, and database. Objective 3: To determine patients’ use of effective contraception in compliance with recommendations in the SmPC for valproate-containing medicinal products, by indication, age group, method of contraception, and database. Objective 4: To determine drug utilisation and prescription patterns over time for alternative medicines prescribed in women who became pregnant, where valproate-containing medicinal products had previously been prescribed or discontinued, by indication, by age group and by database. Objective 5: Based on the results of the above, to estimate the effectiveness of the 2018 RMMs for valproates. Methods We performed an observational times series study including all female subjects of childbearing age (aged 12 to 55 years) from the corresponding databases in Denmark (Danish National Registers, DNR), Italy (ARS Tuscany), the Netherlands (PHARMO Database Network), Spain (Base de datos para la Investigación Farmacoepidemiológica en Atención Primaria, BIFAP), and UK (Clinical Practice Research Datalink, CPRD) between 01 January 2010 to 31 December 2020. In Objective 1, the incident use, prevalent use and rate of discontinuation thereof was estimated per month in each data source, in addition to the change in level and trend in use after the implementation of the 2018 EU RMMs, using an interrupted time series (ITS) analysis design. In Objective 2, first, we separately estimated the proportion of valproate users with a record of a pregnancy test within the 90 days i) before and ii) after the date of valproate prescribing or dispensing per month. We estimated the change in level and trend in these proportions after the implementation of the 2018 EU RMMs. Second, we estimated the proportion of valproate users with a record of contraceptive (prescribed or dispensed with a prescription, or identified through medical events or procedures records) in 90-days before the prescription, or prescribed/dispensed during a contraceptive episode. We then estimated the change in level and trend in this proportion after the implementation of the 2018 EU intervention, using an ITS analysis design. In Objective 3, we estimated the incidence of new pregnancies during a period of valproate use per month and the change in level and trend in this rate after the implementation of the 2018 EU intervention, using an ITS analysis design. In Objective 4, we estimated the rates of alternative medication prescriptions/dispensings for the indications epilepsy, bipolar disorder, and migraine among valproate users and the rate of switching from valproate to an alternative medicine per month. Then, the change in trend of switches from valproate to alternative medications before and after the implementation of the 2018 EU intervention was estimated, using an ITS analysis design. In Objective 5, evidence generated from Objectives 1-4, weighed by the strengths and limitations of the analyses, was used to draw conclusions on the effectiveness of the RMMs, per country and across European countries included in the study. Results Objective 1: There were 69,533 valproate users out of a total of 9,699,371 female subjects of childbearing age from the five participating centres during the study period. The median follow-up time of the study population ranged between 3.5-10.0 years and the mean age at the start of follow-up was always ≥30 years in different centres. The monthly incidence rate of valproate use ranged between 0.01-0.47 per 1000 persons months across databases and the prevalence rate ranged between 1.2-7.7 per 1000 female subjects. While the observed rates were similar for DNR, PHARMO, BIFAP and CPRD, the rates of prevalent use were much higher in ARS Tuscany. We observed a statistically significant declining trend in prevalent use of valproates in all countries/regions, for which an ITS analysis could be performed, but no significant decreasing trend in incidence rates after the 2018 RMMs compared to the period before. The monthly rate of valproate discontinuers ranged between 1-8% across all databases, and in no database we observed a significant increase in trend or level of valproate discontinuation after the 2018 intervention compared to time prior. Objective 2: We included 69,533 female valproate users from the five participating centres during the study period, with a median follow-up time between 4.4-11.0 years and the mean age at the start of follow-up ≥34 years. Due to the limited data on pregnancy tests from all databases, modelling of any trend change in proportion of valproate prescriptions or dispensings with an adherent pregnancy test before versus after 2018 RMMs was not possible. The rate of recorded contraceptive coverage at the start of valproate treatment was low across all centres, as only 0.5-23% of valproate prescriptions/dispensings each month were accompanied by a contraceptive prescription in 90-days before, and only between 0.5-25% of new valproate treatment episode had started during contraceptive use. There was no increasing trend in compliant valproate prescriptions/ dispensings with a contraceptive coverage after the 2018 RMMs across the studied databases, and the only increase in level was observed in PHARMO. Objective 3: We included 69,533 female valproate users from the five participating centres during study period, with a median follow-up time between 4.4-11.0 years and the mean age at the start of follow-up ≥34 years. In general, we observed a substantial number of concurrent new valproate prescriptions/dispensings during a pregnancy time window in ARS Tuscany (386 pre- and 40 post 2018 intervention), BIFAP (330 pre and 20 post) and CPRD (204 pre and 56 post), while there were fewer concurrent events in PHARMO (27 pre and 0 post). However, the rates of concurrent events declined for most databases after the 2018 intervention. There was no data on pregnancy counts available from DNR. Objective 4: We included 69,533 female valproate users from the five participating centres during study period, with a median follow-up time between 4.4-11.0 years and the mean age at the start of follow-up ≥34 years. We found an increasing trend in rates of alternative medicine use for epilepsy and bipolar diseases indications of valproates across the study period in most databases (i.e., DNR, ARS Tuscany, PHARMO and CPRD), while the rates for migraine were mostly steady. The monthly rate of switch from a valproate to an alternative medication was similar across all DAPs and ranged between 1-8%. Running an ITS analysis was not possible for most of the included databases due to the low frequency of switching, but there was a significant increase in trend in switching rates from valproates to alternative medicine after the 2018 RMMs in ARS Tuscany. Objective 5: We found a generally declining trend in prevalence rate of valproate use after the 2018 RMMs in almost all databases (Objective 1), but also no increasing trend in compliant valproate prescriptions/dispensings with a contraceptive coverage (Objective 2). There was a substantial number of occurrences of pregnancy events (as the final endpoint) concurrently with valproate exposure across most included databases, but the rates declined after 2018 (Objective 3). Furthermore, we observed a significant increase in switching rates from valproates to alternative medications only in few regions (such as ARS Tuscany) (Objective 4). Noteworthy, these findings should be interpreted in context of the limitations that we faced, such as an inability to investigate some objectives due to limited data availability on pregnancy test or over-the-counter use of some contraceptives, and the occurrence of COVID-19 pandemic, which has shortened and impacted our post-intervention period and limited our ability to run ITS analyses for some objectives and some databases. Conclusions Objective 1: We observed declining trends in prevalent use of valproates after the 2018 RMMs across all databases. However, there were no declining trends in incidence rate of valproates in none of databases. The rate of discontinuation of valproates was not affected by the 2018 RMMs. Objective 2: We found in general low rates of recorded adherent contraceptive coverage with valproate use across all studied regions/countries, and there was no increased trend in compliant valproate prescriptions/dispensings with a contraceptive coverage after the 2018 RMMs compared to time prior. Due to limited data availability, rates of adherent pregnancy tests and the trend change after the intervention could not be studied. Objective 3: Despite the declining rates after the 2018 intervention, high counts and rates of concurrent pregnancy events with a valproate prescription/dispensing were observed across most studied countries/regions. Objective 4: Although the trend in alternative medication use for most indications of valproates (epilepsy and bipolar disorder) was increasing during the study period, the only significant increase in trend in switching rates from valproates to alternative medications after the 2018 RMMs was observed in ARS Tuscany. Objective 5: Based on the findings on various objectives in this study, we can conclude that there was a small impact of the 2018 RMMs on valproate use and prescribing in the studied European countries/regions. Considering the limitations of this study (such as not studying all PPP elements, the included databases had important limitations, and the study period after 2018 intervention was rather short), the results of other currently ongoing studies are needed to have a clearer picture of the appropriate implementation of 2018 RMMs on valproate use in Europe., The research leading to these results was conducted as part of the activities of the EU PE&PV (Pharmacoepidemiology and Pharmacovigilance) Research Network which is a public academic partnership coordinated by the Utrecht University, the Netherlands. The project has received support from the European Medicines Agency under the Framework service contract nr EMA/2018/28/PE. The content of this paper expresses the opinion of the authors and may not be understood or quoted as being made on behalf of or reflecting the position of the European Medicines Agency or one of its committees or working parties.

2.3.Rationale and background Real-world effectiveness data demonstrated that COVID-19 vaccines' protection against severe SARS-CoV-2 infection is high in the short term but wanes over time, also depending on the virus variants. The vaccine effectiveness (VE) can vary depending on the vaccine brand or type (e.g., among mRNA or adenoviral platforms). Mixing brands for the primary vaccination and/or boosters (heterologous vaccination schedules) has been applied in different countries or regions although the effectiveness of heterologous schedules was not fully understood beyond immunogenic clinical data. With its increased ability to elude immunity and cause reinfections, the SARS-CoV-2 Omicron variant became dominant worldwide and led to the highest ever COVID-19 incidence, also in countries with high vaccination coverage, increasing as well hospitalization and severe outcomes cases in paediatrics populations, particularly in the presence of comorbidities. Moreover, only limited real-life data information on VE for children and adolescents in the EU is available. Further evidence about the VE of homologous (use of the same COVID-19 vaccine for the primary vaccination and booster dose) and heterologous (use of different COVID-19 vaccines for the primary vaccination or booster dose) vaccination schedules are needed both in adult ad paediatrics populations to keep fueling regulatory authorities' preparedness in case of urgent decision-making situations. 2.4.Research questions and objectives To investigate the VE and waning of immunity of diverse COVID-19 primary vaccination (1st and 2nd doses) and booster (3rd dose) schedules with Comirnaty (PF), Spikevax (MD), and Vaxzevria (AZ) vaccines in preventing different COVID-19-related disease outcomes. Primary objectives To estimate the VE, and its waning, in adults (>17 years old) and adolescents (12-17 years old), separately, between heterologous and homologous primary vaccinations. To estimate the VE, and its waning, in children (5-14 years old) between homologous primary vaccinations and non-vaccination. To estimate the VE, and its waning, in adults and adolescents with full homologous primary regimen between those with a homologous booster and heterologous booster, separately, compared to those without any booster. To estimate the VE, and its waning, in adults and adolescents with full heterologous primary regimen between those with any booster and those without any booster. For patients free of prior COVID-19 infection (all analysis), that VE was estimated: By vaccine brand of the primary homologous scheme, the combinations in the heterologous scheme, and booster dose (3rd dose) By age categories. By time since a complete primary vaccination regimen (2nd dose receipt) or booster among the compared groups. Among clinical subgroups associated with a high risk of severe COVID-19 (immunocompromised patients and patients with cancer, transplants, severe renal disease, and Down syndrome). For patients with prior COVID-19, the overall VE of different vaccination schemes against severe COVID-19 and COVID-19-related death was estimated. Secondary objective To estimate the VE against all-cause mortality in ≥60 years old adults with a full primary regimen (homologous or heterologous) between those with any booster and those without any booster. This estimation complements the results of COVID-19-related death of the primary objective. 2.5. Study design Herein, we present a retrospective cohort study to estimate the VE of different COVID-19 vaccines schemes, and their waning, using different SARS-CoV-2 infection-related outcomes: (i) non-severe COVID-19, (ii) severe COVID-19, and (ii) COVID-19 with death. The study used data from 6 European different data sources and focused on the period ranging from the beginning of the vaccination campaign (December 2020) to the last data available from the participating data sources (ranging from December 2021 to February 2022). Thus, it mainly covered the Delta-Omicron predominant SARS-CoV-2 virus variant periods within the full vaccination regimen (first vaccination scheme and booster doses). 2.6.Setting We retrospectively used data from 6 electronic health care databases in Southern, Northern, and Western Europe: the Italian Caserta local health database (IT-INSPIRE srl), the Italian Societa Servizi Informatici (IT-PEDIANET) database, the Spanish Pharmacoepidemiological Research Database for Public Health System (ES-BIFAP), the Spanish Sistema d’Informació per el Desenvolupament de la Investigació en Atenció Primària (ES-SIDIAP) database, the Dutch PHARMO Database Network (NL-PHARMO), and the British Clinical Practice Research Datalink (UK-CPRD) Aurum. Participants were matched 1:1 on the calendar date of the vaccination of interest (2nd vaccination-time0, for the primary vaccination scheme analysis; 3rd vaccination-booster-time0, for the booster vaccination analysis), the calendar date of the 1st vaccine dose, the vaccine brand of the 1st dose (Comirnaty, Spikevax, Vaxzervria), vaccinees age, sex, geographical region, clinical subgroup, and SARS-CoV-2 infection prior 1st vaccination dose. For primary vaccination schedules, the date of cohort entry (time0) was the date when the participant received the 2nd dose. For the booster vaccinations, the date of cohort entry (booster-time0) was the date when the participant received the 3rd dose. For the non-boosted comparators, the same calendar date as the corresponding boosted-matched individual (booster-time0) was used for comparison. Children and pre-adolescents were defined as “not vaccinated” until the date of the receipt of the 1st COVID-19 vaccine dose, thus, potentially selected as unvaccinated control. Participants were considered to have a complete primary vaccination regimen when the record of a 2nd COVID-19 vaccine dose existed after more than 19 days from the 1st dose. Individuals were defined as “boosted” (homologous or heterologous) from the date of the 3rd COVID-19 vaccine dose receipt if at least 28 days after the 2nd one. Participants were defined as “non-boosted” until the date of 3rd vaccine dose administration, thus, potentially selected as non-boosted control. Among individuals with complete vaccination schemes, both homologous/heterologous primary vaccinations and booster/non-boosted cohorts were identified separately. Statistical data analyses Inverse probability weighted (IPW) Cox models (CI, 95%) have been used to derive the average hazard ratio (HR) of COVID-19-related outcomes. The adjusted VE (%) values were estimated as 1 minus the adjusted HR multiplied by 100. Various covariates have been considered potential confounders for the IPW. The VE for each matched cohort was estimated by (i) vaccine brands, (ii) time after vaccination, (iii) age categories, (iv) high-risk of severe COVID-19 clinical subpopulations (for the participant databases able to identify them through diagnosis or medications prescriptions, (v) and SARS-CoV-2 infection before vaccination, and (vi) dominant SARS-CoV-2 variants (defined as the variant reaching 50% of the total sequenced specimens at (booster-)time 0). Three main SARS-CoV-2 variants’ periods have been identified (pre-Delta, Delta, and Omicrons) specifically for each participating country. Random-effects meta-analyses, using the main estimates against severe COVID-19 outcomes from each data source, were performed for clinical subgroups for both adult and children populations. Sensitivity analysis restricting to patients with prior SARS-CoV-2 negative tests was performed to control for surveillance bias. 2.7.Subjects and study size Study Population The source population comprises all children, adolescents, and adults registered in any of the data sources during the study period (December 2020 -February 2022 for ES-BIFAP and IT-INSPIRE and -December 2021 for the other data sources). Eligible study participants had at least 2 years of available healthcare data. The vaccinated study population includes all individuals with at least two recorded vaccinations since the start of the study period. Study size The study cohorts contained more than 20 million adolescents and adults with a complete primary vaccination scheme (1st and 2nd dose receipt) and 3 to 6 months of follow-up across all the participating data sources. We could match ≈24-51% of the population for the heterologous vaccination scheme (comparator) and ≈0.5-1.5% for the homologous scheme. The majority of adults and adolescents were free of SARS-CoV-2 infection prior to vaccination (58-95% across the total matched population for all the data sources). During the study period, approximately 308,000 children with around 3 months of follow-up were vaccinated with two doses and included in the study across all the participating data sources. Based on the matching criteria, a total of 295,573 children were matched for the primary vaccination schedule (95% of the total children). Of the matched children, the main population (97%, 287,050 children) did not have encountered prior SARS-CoV-2 infection. 2.8.Variables and data sources Data sources captured vaccination and outcomes from hospitalization and/or general practice and/or test registries. This study considered different outcomes related to COVID-19: non-severe SARS-CoV-2 infection, hospitalized COVID-19 (severe), COVID-19 with death, and all-cause mortality (secondary objective). The main exposure of interest was the receipt of a different primary regimen or booster COVID-19 vaccine (receiving the same 1st dose brand), the dose, and its brand. Selection of covariates, primary care physician’ visits, clinical conditions, and medication use (including influenza vaccination among others), were collected up to 2 years before (booster-)time 0 (or 7 days before, for visits to control by healthy vaccinees effect) and considered as potential confounders for the IPW. 2.9.Results The shown VE percentages are statistically significant and reported in ranges of values across the data sources, otherwise, the mention of non-statistically significance is specified. Primary Objectives Adults, Primary Vaccination Among 89,528 adults’ matched pairs, overall, homologous primary vaccinations showed a slightly decreased VE (-27% to –36% by data source) compared to heterologous regimens against non-severe COVID-19, mostly receiving AZ as the first dose (VE ranged from -43% to –27% across data sources). By time since vaccination, or, by age categories, no clear patterns were found. In ES-BIFAP, the lower VE with homologous was more marked during the Delta (-39%) than the Omicron (–24%) periods. No differences between homologous and heterologous regimens were observed for severe COVID-19 (estimated in Spanish data sources) and no sufficient cases of death with COVID-19 were found to analyse. Adolescents, Primary Vaccination We matched 1,329 pairs among adolescents. Considering non-severe COVID-19, no differences were found in the VE estimates comparing homologous versus heterologous primary vaccinations. The small sample size hampered the VE estimation related to the other severe outcomes. Children, Primary Vaccination 287,000 children without prior COVID-19 were matched. Considering non-severe COVID-19, homologous primary two doses of both the mRNA vaccines (PF or MD) showed VE, varying from 29% to 77% across data sources, during the Delta predominant variant period, when compared to unvaccinated individuals. VE remained 4-5 months. During the Omicron predominance, VE decreased from 77% to 42% in IT-INSPIRE and reverted to an increased risk of non-severe SARS-CoV-2 infection, from 29 to -44%, in ES-BIFAP. Estimates did not show protection among children with prior SARS-CoV-2 infection. The protection against severe COVID-19 was >90% in ES-SIDIAP (during the Delta period) and ≈50% in ES-BIFAP for PF vaccine versus unvaccinated individuals. In ES-BIFAP, VE during the Delta period was 61%, but non-statistically significant, and 50% during the Omicron period. No data for waning of immunity, from other data sources, vaccine brands and COVID-19 with death were available. Adults, Booster Vaccination 5.6 million adults without prior SARS-CoV-2 infection were matched. 79,076 cases of non-severe COVID-19 among boosted versus 138,638 among unboosted adults were captured in 5 data sources. VE against non-severe COVID-19 ranged 31-69% for homologous boosters and 42-70% for heterologous boosters across data sources, independently from the vaccine brand. Considering severe COVID-19, 1,015 cases among boosted versus 3,362 among comparators were captured in 3 data sources with hospitalization information (mostly in ES-BIFAP and ES-SIDIAP and a few in IT-INSPIRE). Against severe COVID-19, heterologous boosters (homologous doses 1 and 2), independently from the vaccine brand, showed a VE of 73-81% across data sources whereas homologous boosters have a VE of 42-67%, compared to their respective unboosted controls. Considering death with COVID-19, 313 cases of death with COVID-19 among adults who received any booster versus 1,367 among comparators were captured (mostly in ES-BIFAP and ES-SIDIAP and a few in UK-CPRD). Protection against death with COVID-19 was similar among homologous and heterologous schemes (70-88% across schemes and data sources). Duration of immunization varied from 1 to 6 months across data sources and events, independently of the booster schedule. Considering both, severe COVID-19 and death outcomes, no clear VE differences were identified during both the Delta and Omicron periods across data sources. In patients with cancer, effectiveness against severe COVID-19 ranged from 54% to 77% with heterologous and from 49% to 61% with homologous boosters across data sources, whereas in patients with immunodeficiency VE was between 60-78% with any scheme. Among patients with prior SARS-Cov-2 infection, VE against severe COVID-19 was 69% (ES-BIFAP) and 71% (ES-SIDIAP) for heterologous and 43% (ES-SIDIAP) for homologous 3 doses schemes. VE against death with COVID-19 was 92% (ES-BIFAP) for heterologous and 68% (ES-SIDIAP) for homologous boosters in these patients. Adolescents, Booster Vaccination We matched 17,652 adolescent pairs of which 408 cases of non-severe COVID-19 among boosted versus 936 cases among unboosted individuals were captured in 5 data sources (ES-BIFAP, ES-SIDIAP, IT-INSPIRE, UK-CPRD and NL-PHARMO). Among adolescents with a homologous primary vaccination, the VE of homologous booster doses against non-severe COVID-19 varied from 35-67% across vaccine brands and data sources, whereas VE of heterologous boosters was 48% in ES-BIFAP (the only data source in which heterologous boosters were found) for PF as 1st and 2nd dose, and MD as 3rd dose, when compared to the respective unboosted controls. During the Delta predominance period, VE was only observed in Italy (69%) for homologous boosters. During the Omicron predominance, VE varied from 67% (IT-INSPIRE) to 44% (ES-BIFAP) for homologous boosters whereas, for the heterologous ones, was 51% (ES-BIFAP). VE for the homologous boosters lasted up to one month in Italy (75%; later on, severe COVID-19 and death with COVID-19 outcomes due to the low number of cases (Sensitivity Analysis Balancing by any prior testing, the VE against non-severe infection remained 57-59% for primary vaccination among children and 30-55% for boosters in adults. Against severe COVID-19, VE remained moderate (55-59%) for homologous boosters and high (70-81%) for heterologous boosters. Against death with COVID-19, VE was 67-79% for homologous and 77-81% for heterologous boosters among adults. Meta Analysis In adults, the pooled VE of homologous boosters against severe COVID-19 was 62% (95% CI: 57 to 67%; I2=0%) among subjects with immunodeficiency, 54% (95% CI: 41 to 64%; I2=18%) among patients with cancer, 24% (95% CI: -54 to 63%; I2=0%) among patients with a transplant and 57% (95% CI: -20 to 84%; I2=65%) among those with severe renal disease. Additionally, the pooled VE of homologous booster against death with COVID-19 was 73% (95% CI: 63 to 80%; I2=15%) among immunocompromised patients, 75% (95% CI: 65 to 82%; I2=0%) among patients with cancer, and 75% (95% CI: -38 to 96%; I2=63%) for those with severe renal disease. The pooled VE of heterologous boosters against severe COVID-19 (homologous primary vaccination) was 72% (95% CI: 66 to 77%; I2 =0%) among adults with immunodeficiency and 68% (95% CI: 36 to 84%; I2 =77%) for adults with cancer. In addition, the pooled VE of heterologous boosters against death with COVID-19 was 80% (95% CI: 70 to 86%; I2 =0%) among immunocompromised patients and 81% (CI=95% CI: 70 to 89%; I2 =0%) among those with cancer. Among children, the pooled VE of primary vaccination against severe COVID-19 in the Delta predominance period, was 82% (95% CI: -10 to 97%; I2=62%) in Spain. Secondary Objective (any cause of death) In patients aged ≥60 years old, the VE against any cause of death for any booster dose (whether homologous or heterologous), following a homologous primary vaccination schedule, ranged between 72% and 96% across 10-by-10 age categories and data sources (ES, UK and NL). VE was higher during the Delta (75% for homologous and 83% for heterologous boosters) than the Omicron variant period (67% for both booster types). A few heterologous primary vaccinations were used, showing effectiveness (74%) only in 80+ years old in UK. 2.10.Discussion Using real-world data from 6 different databases in Europe, various schemes of COVID-19 vaccination (primary vaccination and booster doses) were identified among adults, adolescents, and children (only in 3 countries) in Spain, Italy, the Netherlands, and the UK from the beginning of the vaccination campaign to December 2021 or February 2022 (in Italy and Spain). Most of the adults participating in our study were fully vaccinated from mid of 202, The research leading to these results was conducted as part of the activities of the EU PE&PV (Pharmacoepidemiology and Pharmacovigilance) Research Network (led by Utrecht University) with collaboration from the Vaccine Monitoring Collaboration for Europe network (VAC4EU). This study was funded by the European Medicines Agency. The views expressed in this article are the personal views of the author(s) and may not be understood or quoted as being made on behalf of or reflecting the position of the European Medicines Agency or one of its committees or working parties. The project has received support from the European Medicines Agency under the Framework service contract nr EMA/2020/46/TDA/L5.06.