Guido Rasi (EMA Executive Director), Peter Arlett (EMA) and Sabine Strauss (Dutch agency) have just published a commentary in the journal Clinical Pharmacology & Therapeutics (available under Arlett’s name at: https://www.ema.europa.eu/en/news-events/publications/scientific-publications#2019-section).
This is an interesting article looking at what they think PV will be in the year 2030. Initially they make note that PV has markedly changed in the last 60 years. It started as a data collection business using serious adverse drug reactions reported as individual case safety reports (ICSRs). It has evolved into an analytic system using new sources of data such as periodic safety update reports risk management plans and “regulatory tools” (not really defined by the authors) and information (“collaborative efforts”) from many stakeholders including the industry, regulators, health care professionals (HCPs), patients and academia. They do not mention lawyers and lawsuits which play a greater role (arguably) in the US compared to Europe.
They then go on to make some disparaging comments about people who make predictions about what is in store for the future of PV! They call these people “clairvoyants” who make
“many claims, sometimes wild” ones. They list some of these predictions including the ending of ICSRs and the use of mobile health tools. They note that technology is advancing, that data accumulation is increasing “logarithmically”, that society is changing, and patients are now engaged in health care decision making. They further note that the goals are not changing: the safe and effective use of medicines to make life better with the minimum of side effects.
Having thus disparaged predictions, they then go on to make three predictions! One is reminded of the Yogi Berra quote: “It’s tough to make predictions, especially about the future.” One is also reminded of some spectacularly wrong predictions: “Heavier-than-air flying machines are impossible” Lord Kelvin and “The cinema is little more than a fad.” Charlie Chaplin.
Nevertheless, the authors then go on to make some predictions:
1. Data collection and reporting of ICSRs will get smarter
They note that although there are 14 million cases in the EudraVigilance database and over 20 million in the WHO database, these only represent a “snapshot in time”. Even so, they maintain that the proportion of safety issues identified from ICSRs is about 55% and product withdrawals also high. Thus they conclude that will not disappear and will remain very useful in safety work. They feel that the ICSRs will evolve by 2030 especially due to e-Health apps and a better collaboration between industry and the regulators. They also feel the bigger and reformulated ICH will promote a more global and better harmonized use and interpretation of ICSRs and signals. They correctly warn that any new systems using robots or artificial intelligence must be carefully monitored and must remain evidence-based. They note that ICSRs are most useful in generating signals and that electronic health records (EHR) are more useful for evaluating already detected signals.
2. Measurement of on-market performance of drugs will develop
They note that PV and healthcare in general have moved from reactive to “predictive, preventive, personalized and participatory (“4Ps”). PV has moved from receiving ICSRs (SAEs) after the fact to proactive risk management activity starting early in the development of a drug. No longer do we just wait for spontaneous safety reports. New sources in real-time such as EHRs, claims data, social media, registries and social media are now playing a role.
They feel that the separation of safety evaluation as pre-marketing (clinical trial) safety and post-marketing safety will disappear and that this will be more of a continuum) where planning is done early on and continuously (My words, not theirs.) Large amounts of data will be received electronically and used by health agencies and pharma companies for product labeling, assessments of “value”, reimbursement decisions and to aid in decision-making at the individual patient and HCP level. They note that the US Sentinel program and the EU Observational Health Data Sciences and Informatics (OHDSI) network are already doing this. They believe that by 2030 monitoring in real-time for marketed products will be done to optimize the safety and efficacy of medicines.
3. Improved Engagement of Patients and Healthcare Professionals
The last prediction is not surprising. The health agencies will work with patients and HCPs “to maximize the positive impact of PV”. This is perhaps optimistic as the authors note that sometimes a health agency warning (e.g. risks of pregnancy when taking valproate) makes minimal or no impact. How this will be done remains to be seen.
The authors’ conclusion to this article is that PV will change rapidly and that measurements of the impact of regulators’ work must be done to ensure that new changes, methods and programs represent improvements that are evidence-based.
Comments
There is not much that is surprising here nor are specifics noted. It is largely more of the same. Surprisingly, little mention is made of academia in the medical, nursing and pharmacy schools. There’s no comment on epidemiology though this is presumably included in the big data comments that they make. There is no mention of the use of Facebook or other such social media to spread messages rapidly or collect data. One could imagine perhaps a negotiation with some of the major social media sites to send out a “blast” when there is a major problem or safety issue with medications or even to send out a call for a specific SAE for a specific drug.
There is no mention of potential downsides that may occur with changes in PV practices. For example, the rapid uptake and use of cannabis in many forms both for recreation and for medical purposes in the US, Europe and elsewhere may present major PV problems with drug interactions and other serious events.
More and more drugs are being “approved” either fully or partially (e.g. “right to try”) with minimal data. This poses a real challenge in PV where the safety profile of a drug may be unknown or minimally known in particular patient populations compared to drugs approved in the current, classic manner.
Other areas that cry out for some new PV methodology include the problem of counterfeits and fake medicines. Not addressed either are issues with inspections and audits both of manufacturing and of clinical and post-marketing use. No mention of hacking or false and incorrect information in the big datasets.
The spread and globalization of drug safety and PV is taken by many if not most observers to be a good thing. It is not clear that this is correct. The level of skill and expertise in the art and science of drug safety is quite variable and the methodology, in spite of CIOMS and ICH and other initiatives, is not well harmonized. Not all regulators arrive at the same conclusions even when looking at identical data. The old saying that “too many chefs spoil the broth” may be correct. Whether 100 plus agencies should look at the same data is a question that has not been addressed in globalization discussions. Duplication of work and validation of signal workups (for example) is appropriate but too much will produce delay and wasting of limited resources. One may wonder whether ICH in its current and new incarnation is as useful as the earlier system was which was largely limited to the major players.
The views in this paper are largely those of the regulators. It is not clear that industry and consumers/patients would fully agree. It is not clear that real-time analysis resulting in the realization that a drug or indication should be withdrawn based primarily on new efficacy data will be accepted. That is, the safety data has not changed but the clinical trial efficacy has not panned out as well in the post-marketing arena. There will be, in such situations, some patients or subgroups who have had perceived efficacy with the drug and resist its withdrawal. The pharma companies too will have a strong opinion in these matters.
In summary, the predictions here are fine, benign and fairly obvious. It would be nice to see a more detailed look at the major trends and ideas analyzing the positives and the negatives and basing predictions on these analyses from other stakeholders.
