The Safety of Medicines in the 21st Century – some thoughts
Sadly, although there are plans to continue the Erice meetings to consider other globally important aspects of pharmacology, therapeutics and drug safety, "A New Erice Report Considering the Safety of Medicines in the 21st Century"1was published just after the death of Professor Giampaolo Velo. He wasthe originator, and champion, of the idea formeetings of global experts- in an environment conducive to both broad yet deep concentration on important issues. This personal view reflects some of my own thoughts taken from the meeting.
I have a feeling of helplessness and failure sometimes looking back over the 50 years of global pharmacovigilance. From a public health and regulatory perspective things have improved, but perhaps not enough, and communication still remains a major issue. Do patients and their health care practitioners have enough useful information to perform their ever increasingly complex tasks in treating patients? Does internet and ‘big data’ help us? How can we better understand the relative risks and benefits of different medicines? Can we prevent the harm from medicines better? What changes in society may help us? These issues and more are considered in the Erice Report
Over the years there has been a steady, but not rapid, improvement in the information in summaries of product characteristics (SPC) and patient information leaflets (PIL), regulatory bulletins and sponsored and specialist publications. Much can be said on this topic, but in relation to safety three issues must be addressed.
- Secrecy and conflicts of interest are still of concern in regulatory and pharmaceutical industry work and publications. This really must stop since other stakeholder users of such information are not privy to the work these groups do, and critical peer revue is not subject to open scientific scrutiny. There is too much data that is not authoritative &/or assessable for interested parties.
- Those unfortunate individuals who have adverse effects are not the norm. Evidence from studies, however, adresses the average or norm and is likely to include only the relatively common adverse effects. Statistical significance is of little or no value in judging whether an adverse effect has occurred in minority groups or not nor can it alone determine causation. Epidemiological evidence alone is valuable in making judgements about frequency of associations only. Other evidence on causation of adverse effects such as considered by Sir Austin Bradford Hill2must be used more frequently as the cornerstone of judgements. Evidence from laboratory experiments (in vivo and in vitro) help us see and understand mechanisms. The root-cause reasons of medication errors are also important as patients potential reasons for adverse effect. The varying dispositions and inter-current illnesses of patients will have also influence whether adverse occur or not effects. More detailed information than we normally collect in pharmacovigilance is needed to understand adverse effects from medicines and their impact on patient and society as a whole
- Patients and their health care practitioners must be seen as the main target for information to which they have a responsibility to contribute. They must understand and take such responsibility for reporting adverse effects. They must be brought in as important partners in pharmacovigilance. The pressures on health care mean that time is precious for practitioners but we must try to find time for such a critical task: not an easy matter.3
The complexity of therapeutics increases continuously and rapidly. Treatments of different kinds and natures are used together and can not only interact for benefit but also risk. The use of different treatment modalities can be very confusing regarding causation and outcome. Pharmacovigilance has shown its effectiveness in seeing early warnings for harm from drugs. Is it not a good idea to extend that vigilance to other treatment outcomes, and so make it easier to assess the overall benefit and harm and thereby improve therapeutic options overall? This balance between benefit and risk of medicines is what patients and their health care professionals need to Medicine4
In any case pharmacovigilance needs to broaden its considerations of harm to considering medication errors and fraudulent and substandard drugs, just as it has broadened to consider interactions with drugs, other chemicals, foods as well as addiction and withdrawal effects. To properly help patients we must move beyond pharmacology to considering all of their clinical care.
Does internet and ‘big data’ help us?
It certainly can but more data, particularly when we do not know its provenance fully or the circumstance and aims for which it was first collected and stored, can be confusing and misleading. The definitions and the categorisations used for data need to be fully understood if we are to be able to use it generally. Combining data from different sources adds to the complexity of understanding its value.
For pharmacovigilance purposes the availability of large data sets have distinct advantages because of the rarity of adverse effects, but more importantly because there is other data on patients which allows for the possibility to study patterns that might give more light on the reason why these few patients and not others suffer the effects.
One unexpected drawback is how large the data set must be to have enough exposed individuals. Longitudinal patient data from over 10 million people in the UK was limited find enough exposed information to analyse even for common drugs.5,6
Another issue is that few longitudinal patient data sets are close to collecting lifelong data outside countries with publicly funded health care. Also most are geared towards collected information on costs and management issues and not on clinically relevant matters.
Internet data is being explored, but so far the issues of terminology prove challenging but this may change with the development of natural language analysis.7
Relative benefit and risk analysis
What is needed by patients and clinicians is succinct information about the best medicines to use in any given situation. The risks and benefits expected need to compared to make good clinical decisions. In the UK the National Institute for Clinical Excellence (NICE) is doing a good job in considering not only benefit and risk but also cost.
NICE is a large organisation and has good resources for this huge task. It provides a service to health professionals globally and is widely respected for the way in which it approaches the issues of benefit-risk-cost. The methodologies it uses, however, cannot cover all clinical situations in providing reliable evidence beyond clinical trials, and quality of life scores (QUALYs), that are transparent, may not be easily generalisable to individual therapeutic situations. Not surprisingly NICE deliberations are lengthy and complex, which is another cause for concern because of the time taken to decisions and guidelines.
The Uppsala Monitoring Centre and others have been working to find methodologies that can allow the use of all kinds of data to be considered in a benefit to risk evaluation and in a transparent way.4Transparency that allows patients to consider the weight that is put on the segment of evidence in those considerations so that they can be aligned with needs, aims and requirements. A fully transparent system has the advantage of modifiability as more and newer data becomes available, thus being able to change clinical and public health decisions in a rational way.
Whatever method for effectiveness and risk is considered a major challenge is how to bring the relevant data together. This is currently a capacity challenge that needs to be tackled.
Prevention of harm
This must also be one of our key future aims. As mentioned above, the use of the internet to assemble big data, and thereby large amounts of useful information, around the occurrence of a harmful effect is essential to understanding the harm. Such understanding of key data and how it can be used to create a picture of what is happening with a medicine is the basis of prevention of adverse reactions.8,9
We must consider prevention as an end-point in pharmacovigilance in a much broader and active way than we do now. Much more emphasis should be placed on capturing the details surrounding an adverse effect, particularly the reporters’ views concerning the factors that might have influenced causation. We need to consider root-causes more often than we do. To achieve all this we will need a significantly better dialogue than we have with health care professionals and patients to give the extra time to finding and sharing such information. Moreover, we will need a consideration about what will constitute a ‘signal for possible prevention’.
Changes in society
I believe we need to foster a more, ‘Let’s join together and find the best ways to treat illnesses’ feeling in society. There is still a strong need to attribute blame when there are suspected adverse effects from therapy: we must move to, not just a ‘no-blame culture’, but instead a culture that sees that poor outcomes are to be actively avoided if possible by changing systems as necessary and mitigated where necessary as soon as possible, through active patients follow up of therapies known to carry serious risks. This should be a primarily public driven exercise with regulation, not driving what patients will put up with, but responding to patient representations for effectiveness and risk balance acceptability and also other ethics issues that include systematic medication errors.
The public as individual and joint voices must be heard and actively encouraged. Adverse effects of medicines do not affect the majority of patients, but the voices of those unfortunate to be affected must be heard and they themselves should feel responsible to provide information to help prevent problems in the future..
Uncertainty about evidence remains , and will remain, a challenge to specialists and lay people in all of life. This challenge may seem to be lessened by sound evidence: it is not. The more we know the more complex the world is shown to be.10Examining this complexity in order to make best decisions is a task worsened by poor and irrelevant evidence, and one of our additional tasks in pharmacovigilance is to make this clear to all stakeholders, not least regulators, the legal profession, and the media. Our duty to the public as scientists and clinicians is to be open and trustworthy about probabilities as they evolve of likely benefits and risks in therapeutics, in order to keep their trust.
One important societal issue that blocks progress over obtaining better evidence is the current conservative and unreasonable approach to confidentiality
Good governance in the provision of medicines and the use of confidential information about people must be instigated and thoroughly maintained, both in principle and practice.
The literal bottom line should be that no evidence about harm and benefit from therapies should be kept secret. The public and health professionals should be allowed to feel responsible in the dialogue about medicines safety and not just bystanders: more detailed information to obtain better evidence is a continuous task.
- Edwards IR. A new Erice report considering the safety of medicines in the 21st century. Drug Saf 2017;40:845.
- Hill Austin Bradford. Proceedings of the Royal Society of Medicine. 1965;58:295-300.
- McCartney M. Tick tock: GPs are ruled by the clock BMJ 2017;385:3955.
- Caster O. Detecting drug risks and weighing them against benefit. Statistical and decision-analytical approaches. [Thesis of Philosophy]. Stockholm: Stockholm University, 2011.
- Norén G, Hopstadius J, Bate A, Edwards IR. Safety surveillance of longitudinal databases: methodological considerations. Pharmacoepidemiol Drug Saf 2011;20:714-7.
- Norén G, Edwards R. Opportunities and challenges of adverse drug reaction surveillance in electronic patient records. Pharmacovigilance Review 2010;4:17-20.
- Antoni F, Wisniewski Z, et al. Good signal detection practices: evidence from IMI PROTECT. Drug Saf 2016;39:469-90.
- Aronson J, Ferner R. Joining the DoTS: new approach to classifying adverse drug reactions. BMJ. 2003;327:1222-5.
- Ferner R, Aronson J. EIDOS: a mechanistic classification of adverse drug effects. Drug Saf 2010;33:15-23.
- Rocca E, Anjum R, Mumford S. Causal insights from failure post-marketing risk assessment of drugs as a way to uncover causal mechanisms. https://www.nmbu.no/download/file/fid/32376 (accessed 04 Oct 2018)
I Ralph Edwards
Uppsala Monitoring Centre, WHO Collaborating Centre for International Drug Monitoring, Uppsala, Sweden