We requested from Eurostat detailed information about deaths caused by adverse drug reactions as part of research performed for our previous blog about the number of lives saved by pharmacovigilance activities. Saving lives and reducing healthcare costs by decreasing the number of deaths and hospitalization caused by adverse drug reactions was used by the European Commission as one of the reasons to update the EU pharmacovigilance legislation.

In the past 10 years since the legislative updates, the cost of pharmacovigilance activities has increased significantly but there is no published data on the measured impact on the death and hospitalization rates.

We have been missing an analysis by the EMA on effectiveness of their processes and the EU pharmacovigilance system as a whole. While we still need to wait for this, some conclusions can be made based on the publicly available data.

Review of the Eurostat data further strengthens our suspicion that EU pharmacovigilance activities have not been effective and need an update.

The Background

As presented in our blog, in 2008 European Commission estimated that approximately 200,000 deaths were caused by adverse events from medicinal products annually. Adverse drug reactions were estimated to be responsible for about 3 – 10 % of all admissions to hospitals in the EU. Out of all hospitalized patients 2.1 – 6.5 % suffered an ADR and 0.12 – 0.22 % of hospital admissions resulted in death due to an ADR.

Our research revealed that no information has been published by the EMA related to the impact of the updated legislation and the improved pharmacovigilance system. We looked at the data in EudraVigilance which showed that while the number of reported individual case safety reports has increased, the portion of death and hospitalization has remained the same.

As an example, we also looked at the statistics in Finland, which showed that less than 1 person per year has been reported to have died due to an adverse event caused by a medicinal product between 2010 – 2019. A different category, poisoning by certain medicinal product categories, had larger figures totalling about 250 deaths per year in the country.

We contacted Eurostat to receive more detailed information about the number of deaths caused by ADRs in the whole legislative region. We had to contact the customer service because the death categories for adverse drug reactions and overdoses (poisoning) had been grouped together with other external causes of death.

The Data from Eurostat

We have analyzed the information from Eurostat for the years 2010 to 2018.

The data collection has been mandatory in the EU since 2011 under Commission Regulation (EU) No 328/2011, which means that the data for 2010 was still from a voluntary data collection. The numbers are still pending for France for the years 2017 and 2018. We requested the data coded with ICD-10 but Greece has used the old ICD-9 coding until as late as 2013.

The data extraction was anonymised to ensure statistical confidentiality. Eurostat applies the k-anonymity with k=4, i.e. they anonymise the underlying cause(s) of death for any given combination of the location, sex, year and age group if there are 1 to 3 deaths. The anonymised records are under the category ICD-A-R_V-Y and contain the total number of deaths. Eurostat cannot provide any information if the category actually includes any death from the adverse drug reactions. Due to the characteristics of the data, this approach anonymises the causes of death for age groups with a low death rate in the countries with small populations.

In practice, if analysing data for rare causes of death, it is very likely that most of the lines with ICD = A-R_V-Y do not contain any cause of death for which this extraction had been done.

Data Analysis

Deaths reported to have been caused by adverse events

Deaths caused by adverse events or by poisoning by medicinal product

Deaths caused by ADRs, poisoning and anonymized causes

Conclusion

The annual number of deaths caused by ADRs and poisoning related to medicinal products has not decreased since the 2010. In fact, the number has steadily increased.

Based on the Eurostat statistics, the European Commission original estimate for number of deaths caused by ADRs seems to be grossly wrong.

Obviously, deaths are only one thing to look at when measuring the effectiveness of PV activities. As we showed in our previous blog, another important measurement is the number of hospital visits and their causes. Further research may be warranted.

For a full analysis and complete measurement of effectiveness, a review of the number of individual safety reports reported to EudraVigilance is required. A key subsection of this data is the number of safety reports received from the patients. While it is still best practice that the patient discusses the adverse events they experience with their treating physician, it is great that the patients can have more control over their own treatment and that this is encouraged by the national competent authorities.

We are hoping to soon see updates from the EMA to measure the effectiveness of their processes. It would be especially interesting to see this regarding the signal management process which is currently in a pilot phase or related to the update of electronic reporting requirements of 2017.