Pharmacovigilance: science or bureaucracy?

What are we talking about?

To get started it is worth looking at the ways in which “bureaucracy” is used. According to Google definitions bureaucracy has 4 meanings:

1. System of government in which most of the important decisions are taken by state officials rather than by elected representatives.
2. State or organization governed or managed as a bureaucracy.
3. The officials in a bureaucracy, considered as a group or hierarchy.
4. Excessively complicated administrative procedure.

The reason for starting this way is to point out that a bureaucracy is concerned with administration and intended to ensure that all functions follow rules. A similar term is “governance”. It is the last definition that reminds us that such systems can get too heavy and tie us all up in detailed rules about everything, and those rules may not have been worked out wisely. In essence a “standard operating procedure (SoP)” is one of the sets of rules that make up a bureaucracy, and the aim of an SoP is that if anyone follows it they will get the desired, same, result every time. If, however, the result is poorly defined no amount of good governance will ensure success.
Similarly “science” is defined by Google as:

1. The intellectual and practical activity encompassing the systematic study of the structure and behaviour of the physical and natural world through observation and experiment.
2. A particular area of science.
3. A systematically organized body of knowledge on a particular subject.

The word “systematic” is important in science and that means to operate according to a plan or logic, and, for science to work, the logic is subject to proving or falsifying experimentation. To understand science is also to understand the relationships and structure. The essence of science is change, moving from hypothesis to experimentation; to new knowledge through imputation/deduction; and to new hypotheses. It is an iterative process that may never end.
A simplistic view of bureaucracy versus science is that the former is rule based functioning with an aim of efficiency by removing error, whereas science is concerned with challenging the rules and using experimentation to determine error or deficiency. Bureaucracy is good for the governance to ensure that established best practices are maintained. Science is good at determining what the current best practices are by experimentation.

What is public health?

Public health has a long tradition in developing rules that have been effective in ensuring that all aspects proven necessary for optimum health in a population are adhered to: a bureaucratic exercise. Epidemiology was developed as a basic scientific tool to provide evidence for normative actions for health.
It was broad public health concerns that led to governments overseeing the quality, efficacy and safety of new drugs. Over years, regulation of marketed drugs has developed. The early clinical trials for a new drug (now called phase 1 to 3 studies) need good governance not only for ethical, logistic, and legal reasons but also for scientific ones. The bureaucratic processes that have been developed by ICH and others have allowed us to understand much that is important about the efficacy and hazard that might be associated with a new drug, whilst at the same time preventing unnecessary harm to subjects.
We know more and more about isolated body systems and indeed we try to isolate systems within the body when testing drugs to remove non-target variables and to reduce noise in the experiments.
Reducing noise is very important to start to understand whether a drug will work and at the same time to test a given set of hypotheses about its actions within the body: we hope to have a drug that will work in the way our current knowledge tells us. In use in clinical, non-experimental situations however complexity and interaction between drug , body and disease may produce new findings/outcomes.

What is Pharmacovigilance (PV) now?

The practice of PV has paralleled that of clinical pharmacology and the developments in medicinal products: it is a new and evolving science over about 5 or 6 decades. We know the human body to be hugely complex and that we are not sure about many of the body’s functions and certainly not the details of interactions within the body or with the body and disease entities or introduced chemicals including drugs.
PV starts at the point when a drug is given to humans and has the challenge of answering the question, “Does the drug work in all situations and in all people in the way we expected?” There are also secondary questions, “Why does the drug not work as expected?”, “What can we do about it?” These might be important public health issues, but more often than not they are issues affecting minority groups of patients: sometimes very small groups.
After the Thalidomide disaster pharmacovigilance was an idea first proposed by Prof. David Finney to find new suspicions of harm caused by the drug itself when used in ordinary clinical practice. This was because it was clear that controlled clinical trials of drugs were inadequate in finding all the harm that might occur when drugs are used in clinical practice.
The main aim was to use the reports of doctors who thought their patients were harmed by drugs and bring them all together, so creating a global database. Expert evaluation of the cases, together and singly, using additional data and expert experience in pharmacology and clinical medicine, would result in hypotheses for further action. From the beginning, therefore, the primary motivation for pharmacovigilance was the essence of science - hypotheses.
It was, however, also clear that the governments who had broad responsibility for health care and the registration of safe and effective drugs to be on the market would be very interested in the hypotheses of harm and their further analysis and interpretation for public health. Furthermore, the pharmaceutical industry also has a great interest in the issue of public health consequences of the drug products they market. They have no wish to be responsible for ill health and they bear the consequences of litigation and the payments if negligence is legally proven, whatever the reasons or causes might actually be.
For these reasons it is not surprising that the concentration has been on ensuring that timely warnings about the possible harms from drugs has become the focus of attention, particularly if the numbers of patients harmed seems to be large and the harm serious. Since epidemiology is a major public health tool this is used as pharmacoepidemiology to ‘prove’ risk. It is also not surprising that a bureaucracy has developed around the way these ends are achieved.
Standardisation is a bureaucratic process that is very useful for definitions and checklists for data quality and handling, but analysis of data should remain a flexible process performed by suitably qualified people. The possible bureaucratic reliance on so-called knowledge finding algorithms is reasonable for finding previously defined types of adverse events, either singly or in clusters, in any kind of formatted data and at predefined levels of difference (disproportionality). They do not find adverse events or any other information that is not precisely predefined, so they will not find the new and unexpected – that is for humans! Neither will they find everything because of the pre-set sensitivity limits.
Because of the almost inevitable growth of bureaucracy predicted by the German philosopher and sociologist, Max Weber (see further reading suggestion below), PV has become a public health exercise with extensive developments in bureaucratic support. Both regulatory bodies and the pharmaceutical industry want to demonstrate that they do everything they can to ensure that all reports of harm they receive are collected, filed, collated and result in appropriate warnings.
Current PV has become overwhelmed with bureaucracy because even these latter limited aims are thought to require extensive quality assurances to ensure good governance. One major flaw in this bureaucratic approach and processes is that the primary source of the information, doctors, are not incorporated into the system neither for the quantity nor quality of their reports.

1. There is considerable and variable under-reporting (often quoted as about 95% under-reporting) of suspicions of harms by doctors
2. The information contained in the reports has not often been of sufficient quality for an independent expert to work out why there is a suspicion of causation (what factors where considered in the clinical diagnosis of the patient)

Accepting either of the two statements as true means that the very foundation of PV needs attention, and the current governance system is inadequate if not misplaced. Bureaucracy is conservative and rules based, but the rules in this instance have been applied to the reporting of suspected adverse drug reactions by industry with much expense and resource needs. Would it not be better to make an assessment of the current system and its performance and then try new approaches in a scientific way to get better results? One can think of improving the communication systems for health care practitioners and patients to report, as one major way forward. Some such work is already underway, but how will we know whether changes will make any difference without a scientifically based protocol for change and key success factors being agreed and tested? One thing is certain: a bureaucratic approach will maintain the status quo, which we know has deficiencies.

3. Reliance on pharmacoepidemiology as a way of disproving causation is logically flawed
4. The communication of useful knowledge on the safe and effective use of drugs, both to patients and health care professionals, is not fit for purpose
5. Much harm is caused through medication errors that can occur at any point along the chain of activities - from incorrect diagnosis of the indication for the drug, through to the adherence to the drug by the patient, as well as collateral issues interfering with those activities.

These next three statements are aimed at pointing out that the top down, public health approaches to PV are limiting. For many reported harms, their relative rarity makes them much less likely as causes than other disease entities. The sample sizes of most studies will not have the power to detect rare causes of harm, anyway.
Pharmacoepidemiology is concerned with relative probabilities, and for health professionals to know what those relative probabilities are would be very useful in their diagnostic considerations, as well as in choosing therapy. Moreover, pharmacovigilance reports together with epidemiological information should tell us much more about risk groups, and therefore preventability. There are two major points here: the focus of PV is on individuals or small groups that are not the public health norm, and secondly that the idea of producing a benefit from pharmacovigilance activity is dependent on increasing the knowledge of both patients and health professionals. Simply providing information, as is done in the Summaries of Product Characteristics, is not enough. It is thorough knowledge and understanding that is necessary for good practices in any field. For that to be realised, the use of the best communication science and technology is essential. This is particularly true of a rapidly changing field such as therapeutics One other area where bureaucracy might be very helpful is to ensure that all efforts are made to communicate clinically useful information to health care professionals and patients in the most useful way. There is no current guidance that makes those responsible for providing information to health professionals that also requires that they do their best to impart knowledge and aiding understanding.
Bureaucracy produces more and more information, and information overload reduces the acquisition of knowledge, and critical thinking.

Conclusions

The foundation of science is hypothesis, which was and is a main aim of PV.
Broad, normative public health approaches and epidemiology have had probably had little impact on the safe and effective use of medicines in the individualised decisions made in clinical practice, though we have not really tried to assess that. On the other hand the many drug related problems that occur do have huge public health impact because so many drugs are used.
Standardisation has been valuable to define both content and processes to be used globally, so allowing data to be grouped and performance to be compared.
The main aspect lacking for wise therapeutic decisions - for the most benefit and least risk - is knowledge. Science must be foremost and bureaucracy only very carefully employed to ensure aspects of a governance framework which ensures that all useful information is reliably and readily available.
The best communication support to improve the knowledge of health care professionals in the relative effectiveness and risks of medicines might be a more effective focus of bureaucracy.

Further reading

1. On bureaucracy and pharmacovigilance:
   Hugman, B ‘Protecting the People? Risk communication and the chequered history and performance of bureaucracy’ Drug Saf 2012; 35 (11): 1005-1025 0114-5916/12/0011-1005/$49.95/0
2. On the issues of information versus knowledge and thinking:
   Stiegler, B. trans. Ross, D ‘States of Shock: stupidity and knowledge in the 21st century’ 2015; Polity ISBN 978 - 0 - 7456 - 6494.
3. Also a review of the above book:
   Benson, P ‘States of Shock by Bernard Stiegler’ 2016; Philosophy Now: 111: 43-44.