What is Still Missing – What We Must Do in the Future

WHAT IS STILL MISSING – WHAT WE MUST DO IN THE FUTURE

The pharmaceutical industry is poised on the edge of new opportunities and challenges in the new millennium (Edwards, 2000). Better and faster ways to develop new medicines clearly give one oppor-tunity, but the real excitement is in the area of genetic knowledge and manipulation, which allows unprecedented interference with disease processes. The industry is faced with challenges to become ever more profitable, and this has resulted in what might be called management experiments of re-structuring, merging, outsourcing, virtual companies and so on. There is an aim to market medicinal products globally and fast. Even recreational drugs are a possible legiti-mate consideration for the pharmaceutical industry in the future. All of this has implications for the safety of medicines, and the most obvious issue is that the rapid exposure of large numbers of people to novel products, which might have profound effects for ill as well as good.

Many publications attest to the high propor-tion of hospital admissions that are related todrug injury (Lazarou, Pomeranz and Corey, 1998; Pirmohamed et al., 2004). Most other disease inci-dences do not come close to drug injury as a cause for morbidity. Moreover, it seems likely that about half of these events are avoidable. A chronological examination of the literature on drug-related morbid-ity makes it clear that this public health problem is not decreasing. Why is this?

More drugs become available on the market all the time, and this may itself be a factor in keeping the incidence of drug-related morbidity high. In addition, there can be a higher reporting rate for adverse effects associated with new drugs (the Weber effect). This comes about because of clinical interest in the new drug, the possibility of a novel ADR profile, as well as effects which may have come about because of lack of clinical experience with the agent (e.g. first dose hypotension with calcium antagonists, depen-dence and withdrawal with selective serotonin reup-take inhibitors).

Multiple drug use may result in adverse interac-tions, causing ADRs or lack of efficacy (Meyboom et al., 2000a,b). Not only does polypharmacy occur when a single physician is treating compound disease processes but with increasing specialisation, more than one doctor may be prescribing without another’s knowledge. In addition, the patient may be taking over-the-counter medications and herbal preparations. Treating compound disease also requires considera-tion of the interaction of concomitant disease on drugs used for the target illness. More patients are treated for multiple serious illnesses: elderly patients need specific consideration in this respect, and a larger part of the population of most countries is in the geriatric age group.

Fraudulent drugs may cause problems of lack of efficacy (Meyboom et al., 2000a,b) and issues relat-ing to adverse effects resulting from excipients. This growing problem, which affects developed and devel-oping countries, needs a different approach to phar-macovigilance. Certainly, there are many countries which still need to develop effective drug regulation.

Misdiagnosis, bad prescribing, bad dispensing and other poor practice leads to drug injury, but there may be correctable reasons for this poor performance. It is clear that the pressure is mounting on doctors and other health professionals. The technical and professional complexity of their work is increasing and to this we must add an increasing administrative and bureaucratic load. Undergraduate medical train-ing does not devote sufficient time to drug safety, and post-graduate education is too frequently concerned with the latest therapy and the importance of being up to date in the scholarly rather than practical sense. There is unending pressure on doctors, including the threat of litigation for even the most genuine of errors by the most careful of doctors. Patients are increasingly informed on medical matters and are encouraged, quite rightly, to understand and be active partners in their therapy instead of passive subjects. Unfortunately, the reliability of information sources is very variable, including a huge amount of information accessible to patients on the internet. Increasingly, therefore, doctors are required to justify their advice on therapy and even to undo confusion because of conflicting information.

There may be more reasons why drug-induced injury continues to be a public health problem, but it seems clear that much of it relates to fundamental issues of health professional education and working circumstances. The rest has to do with more drugs, more technical innovation and increasing information overload.

The relationship between clinical practice and patient harm has recently been given a much higher profile. The developments spearheaded in Australia, the United Kingdom and the United States of America have been recognised by a global effort to tackle the problem: the WHO World Alliance for Patient Safety. A central theme in the work will be to understand patient safety and medical error in a systems sense and to avoid a ‘blame culture’ (Edwards, 2005b). Pharma-covigilance must be a part of this effort and some of the steps below need to be considered in this context.

There are five broad activities that are essential to pharmacovigilance. These are

·        suspected ADR signal detection and formation of hypotheses,

·        analysis of all issues around the signal, particularly confirmation (or refutation) of hypotheses, estima- tion of the size of the risk and whether particularly susceptible patients exist,

·        consideration of possible changed effectiveness-to- risk issues in therapy,

·    communication of information to health profes-sionals and patients in a useful way and possible regulatory action and

·    consequence evaluation.

Each of the above steps will be considered below in relationship to some change, critical to make more progress. A basic assumption is that, since drug therapy very rarely constitutes epidemic risks, public health is very much concerned with securing the best benefit–risk for minority groups as well.

DRUG SAFETY SIGNALS

Suspected ADR signals may be related to a new drug or to the way in which any drug is used in the community. Because many hospital admissions are caused by avoidable ADRs, we should take much more notice of reports of known ADRs to older drugs and generally regard any ADR report as something that has concerned a reporter enough to send it! This means not just concentrating on adverse reactions to new drugs (serious and unexpected) but to encour-age health professionals and consumers to report any significant adverse effect relating to drug therapy. We need to provide the right climate for health profes-sionals to be observant and critical in their diagnoses and therapy, so that they do not miss any piece of new information that may make therapy safer. IT and data mining can improve the transfer and analysis of the additional reports, respectively. In addition, it will be necessary to widen the scope of reporting to include adverse reactions to herbal and other tradi-tional remedies, drug misuse, abuse, poisoning and overdose and unexpected lack of effect if we really wish to tackle the public health issues surrounding drug therapy comprehensively.

Multipurpose health databases should be used to monitor drug safety signals much more than they are at present. Such databases should be planned so that appropriate data can be captured. Reports from consumers should be acted upon, both with a response to the individual and to the general public where appropriate. The UMC has recently worked with IMS Health on data mining the latter’s disease analyser data. This has started with a successful pilot project using approximately two million patients’ fully anonymised health care records. The potential to find unknown patterns of links between prescribed drugs and outcomes, even in sub-groups, is great, including some of the challenges raised in the next sections.

SIGNAL ANALYSIS AND IMPACT – HYPOTHESIS TESTING

Very many signals are produced, and our ability to analyse them is limited. Currently, there seems to be little consistency over what signals will be considered further. Serious signals that appear new, and relate to new drugs, usually elicit regulatory action. Less serious signals that may none the less have an impor-tant impact on morbidity, and compliance may not be investigated so rigorously even when the numbers build up. Epidemiological studies may take months to years to perform during which time thousands of patients may be exposed to the signalled risk.

This period of new signal analysis is rarely made transparent, and controversies tend to linger. Almost the whole effort of this vast collection machinery for clinical case report information is directed towards finding new ADR signals. Little use is made of the data for other signal work, such as

·    finding at-risk groups (e.g. do some ADRs occur disproportionately with age?);

·    interactions (do known reactions occur more frequently with certain medicine combina-tions?) and

·    ADRs related to usage (e.g. do certain reactions occur more frequently in certain countries? At higher doses? Are there systematic errors in use?)

This is not surprising, because the quantity of data is so large and most national centres have few resources. Several needs are apparent if we are to meet the chal-lenges of the future. Amongst the most important are

·    to encourage clinicians to report clinically rele-vant experience, including better details of what happened;

·    to do root cause analysis on cases;

·    to give advice about the diagnosis and management of ADRs;

·        to improve the rapid transmission of quality infor-mation to national centres and industry, and thence to the WHO database;

·    to find ways of supporting the examination of large amounts of disparate information and

·    to be able to bridge the gap between a tenta-tive signal from raw ADR data and observational studies that use specific protocols.

BENEFIT TO RISK ANALYSIS

Much of the debate about comparative benefit and risk is bedevilled by failures of logic and definition (e.g. clearly differentiating between ‘harm’ and ‘risk’) and the use of different criteria in different situa-tions. It is very important that these issues are identi-fied in any critical review of information. The UMC developments in this area involve

·    promotion of the principle that responsible safety information must involve an element of benefit– risk analysis, that is what the patients actually feel about responses to therapy. Newer quality of life measurements will aid this process, as will a broader view of the information through consumer and health professional spontaneous reports. They should be seen as consumer concern reports, not as part of epidemiology. Bad quality (having little information) reports should still raise concerns, even though they may not be of much use in determining actual causality between drug and effect. They still expose situations of public perceptions of risk which need to be addressed;

·    the further development of definitions that are acceptable to the WHO collaborating national centres;

·    to develop much further on the CIOMS IV guide-lines on the ‘Principles of benefit-risk compari-son’ and

·    the development or promotion of methods that will enhance more rigorous benefit–risk analysis, for example

·    comparing like with like;

·    the use of best-case and worst-case analysis for uncertain safety information;

·    international analysis to highlight and to determine reasons for differences in reporting of ADRs and

·        analysis of ADR reports for comparator medical products when important safety signals are raised.

COMMUNICATION OF BENEFIT–RISK INFORMATION

Currently, the emphasis of communication is on deciding whether a drug should be available or not and communicating that information and the provision of official information in summaries of product char-acteristics (SPCs) and their equivalents or in formula-ries. Decisions are made by regulators and the industry and their professional advisers as a result of a debate that is not transparent to consumers in most countries. Medicines are somewhat different from most other consumer products, in so far as patients generally do not have the ability, either because of lack of knowledge or insight to make good choices about their own treatment unaided by information presented in a way useful to them and without professional advice. The question then arises as to whether health professionals, as learned intermediaries, have the correct or sufficient information on the benefits and risks of drugs from information that is readily avail-able during clinical practice, for example reference books and SPCs.

Patient information leaflets are now promoted by some authorities, such as the EU and industry. These moves seem reasonable, but there must be a review of their effectiveness.

Communication to health professionals on adverse reactions needs to give some idea of their likelihood, severity and possible outcome to be useful to a clini-cian and, of course, their patients. Little of this infor-mation is made available nor is the level of certainty made clear on the evidence for most reactions.

CONSEQUENCE ANALYSIS

As far as possible, the likely consequences of a response to a safety concern should be considered before the action is undertaken. Input should be sought from experts in communication science, patient groups, practising health professionals and others when trying to predict consequences. This knowledge should guide choices between the options for action available. For example, a consequence analysis should be planned before a warning about a drug is given out or the drug is taken off the market. This analysis should be in two parts: an early investigation designed to ensure that the expected effect was achieved, so that a correction or reinforcement can be applied as necessary, and a later evaluation to ensure that a posi-tive response is maintained. The UMC has previously looked at the way in which the signals it produces have been used in national centres (Edwards and Fucik, 1996, Ståhl et al., 2003).