Part I: BASIC CONCEPTS AND DEFINITIONS
Introduction: The importance of safety to patients and healthcare professionals
All pharmacologically active substances have the potential to cause harm. Medicines are taken by patients with an expectation of benefit and usually with some awareness that side-effects might occur. However, some patients will experience serious harmful effects from medicines, leading to death or permanent disability.
Despite the relative safety of modern medicines and extensive testing requirements, adverse drug reactions (ADRs) are an important cause of morbidity and mortality in the developed world. They are a frequent problem in general practice and they account for around 6.5% of hospital admissions (Pirmohamed et al. 2004). A report from the Audit commision in 2001 suggested that adverse drug reactions and medication errors cost the UK NHS about £0.5 billion per annum. Furthermore, it has been suggested that ADRs are among the top 6 causes of death in the USA (Lazarou et al., 1998). They are certainly the most important form of iatrogenic disease and many serious ADRs are preventable.
A side-effect is an unintended effect of a medicine. Normally is it is undesirable but it could be beneficial (e.g. an anxiolytic effect from a beta-blocker prescribed for hypertension).
An adverse drug reaction is an unintended and noxious effect that is attributable to a medicine when it has been used within the normal range of doses (i.e. it does not include effects caused by overdoses).
An adverse event is an undesirable occurrence that occurs in the context of drug treatment but which may or may not be related to a medicine.
A serious adverse reaction or event is one that is fatal or life-threatening, or leads to hospitalisation or disability.
In individual cases, it is often difficult to determine whether or not a particular event was caused by a medicine. This requires clinical judgement taking into account:
Timing of drug administration and the onset of the possible ADR
Likely alternative causes (e.g. underlying illness)
Relationship to dose and withdrawal of the drug (and re-challenge, if applicable)
Whether such reactions are recognised with this drug or others in the class
Adverse drug reactions usually mimic naturally occurring diseases and it is important to maintain a high level of suspicion that unexplained disease could be drug-induced. Unless they are well-recognised, it is important to report serious suspected adverse reactions to the relevant authorities e.g. in the UK on a yellow card to the MHRA (see below).
The concept of safety
Safety means relative absence of harm or a low level of risk that, in the particular treatment context, can be considered acceptable. There is no such thing as absolute safety.
Safety is a moving ball – there is a need to re-evaluate it as experience accumulates, and treatments previously considered acceptably safe may become "unsafe" in the light of new evidence.
Risk is the probability of an adverse outcome. It may be expressed in the following terms:
Absolute risk (e.g. 1 in 100)
Relative risk (e.g. two-fold increase compared to alternative or no treatment).
Relative risk is easier to measure but absolute risk is what the user needs to know. The inverse of absolute risk is a useful tool known as the number needed to harm (NNH). This is analogous to the number needed treat when benefits are being considered. For example, if the absolute risk is 1 in 100, the NNH is 100.
It is much more difficult to determine that an effect is absent than to measure one that is present. Even if many people have used a drug without experiencing a particular effect, it is impossible to be certain that this will apply to the next person to use it. The rule of three is a simple and useful tool when zero cases have been observed in a defined population. Simply dividing the size of population by 3 approximates a 95% confidence limit for the highest level of risk that is likely to be present. For example:
If 900 patients use a new antibiotic and 0 allergic reactions occur then it is unlikely that such reactions will occur more frequently than 1 in 300 patients (i.e. 1 in 900/3).
There are two components to safety:
Some substances are intrinsically safer than others at therapeutic doses (compare, for example, paracetamol and any cytotoxic drug). In principle, two drugs with similar indications and benefits are unlikely to have major differences in their intrinsic safety (although the types of adverse reactions produced may be different) – such differences would not be acceptable to regulatory authorities.
In practice, therefore, the key issue for clinicians is user-dependent safety – doing everything possible to minimise the possibility of serious adverse reactions (whilst still obtaining the desired therapeutic effect). For prescription medicines, most of the opportunity (and responsibility) for safe use rests with the prescriber although pharmacists can also play an important role.
Since absolute safety is an unattainable goal, the aim is to use medicines with an acceptable level of safety. Various factors need to be considered in judging whether safety is acceptable:
The level of absolute risk and the potential health consequences
The seriousness of the disease for which treatment is given and the level of benefit expected
The risks and benefits of alternative approaches
The perspective of the individual who is to be exposed.
The balance of benefit and risk
There is no magic formula when it comes to deciding whether the benefits of treatment outweigh the risks (or vice-versa) – the process requires a judgement. Ideally, a balance sheet would be constructed analogous to those used in finance and the debits would be subtracted from the credits, hopefully leaving a positive balance. The problem here is that the credits and debits are not usually measurable in the same way and there is often uncertainty about the size of some of the entries. Nevertheless the balance sheet is a helpful analogy – "to achieve these benefits it is reasonable to accept those risks (or not)". Such judgements are made at a population level by regulatory authorities and, for any marketed medicine, the authorities must consider the overall balance to be favourable for the indication which they have licensed, provided it is used after consideration of the user-safety issues listed above. Thus, if a prescriber is using a licensed medicine in such a way, in principle, the balance of benefit and risk and should be positive.
Pharmacovigilance is the process of evaluating and improving the safety of marketed medicines through risk management. The key players are regulatory authorities, pharmaceutical companies and health professionals.
The most important elements of the pharmacovigilance system in the UK are the Yellow Card scheme (run by the Medicines and Healthcare products Regulatory Agency) and Prescription-Event Monitoring (which uses "green forms" and is run by the Drug Safety Research Unit). These are described in part III. Information from these and other sources is used to identify and study adverse reactions and to formulate recommendations for using medicines more safely. Such recommendations are communicated to doctors and pharmacists through letters, bulletins and changes to product information. During the initial post-marketing phase (about 2 years, during which time the drug carries by the black triangle symbol – ▼ – indicating that all suspected ADRs should be reported), it is common for additional warnings or precautions to be added to product information as practical experience of a new drug is gained. In addition, about 4% of all drugs are ultimately withdrawn from the market for safety reasons.
Part II: HOW THE SAFETY OF MEDICINES IS TESTED
As indicated in part I, there is a need to continuously evaluate the safety of medicines. This process starts before humans are exposed, and continues throughout the development and marketing phases. Broadly, the safety of a medicine is usually tested sequentially in four phases:
Animal (pre-clinical) studies
Healthy human volunteer studies (sometimes referred to as Phase I studies)
Clinical trials (Phases II-III)
Post-marketing (Phase IV).
Pre-clinical studies are usually conducted in rodents (rabbit, mouse, rat) and dogs. They aim to establish dosage levels below which toxicity is not observed and to identify the organs adversely affected by higher doses. The most important potential effects studied are:
Major organ toxicity
Mutagenicity (i.e. able to induce genetic mutation)
Teratogenicity (i.e. producing physical defects in the embryo).
Even at this stage, some adverse effects might be acceptable depending on the ultimate target population for the drug. For example, adverse reproductive effects would be considered unimportant for a drug that is to be used exclusively in an elderly population.
Adverse drug reactions (ADRs) may or may be not specific to particular species. When studies in animals demonstrate major toxicity, further drug development is usually precluded and the level of toxicity in humans remains unknown. When no major toxicity is demonstrated in animals, clinical development proceeds but some ADRs are specific to humans (e.g. the multi-system fibrotic reactions caused by the beta-blocker practolol).
Overall, whilst pre-clinical studies are rightly mandatory before humans are exposed, their predictive value for human toxicity is no more than moderate. Thus they provide only limited reassurance that use in humans will be acceptably safe.
Human volunteer studies and clinical trials
For most medicines, the first human exposure takes place in healthy volunteers (cytotoxic drugs are an exception). The purposes are to establish a possible dosage regimen, how the drug is handled and what the effects are on a variety of standard parameters (e.g. pulse and blood pressure, ECG, haematology etc.). Assuming the drug appears to have no major untoward effects, it can then be studied in clinical trials which include patients with the target disease(s).
Clinical trials are usually designed to study both safety and efficacy. With regard to safety, all adverse events occurring to the patient after exposure to the drug and a comparator (which may be a placebo or an alternative drug) are systematically recorded. The data are analysed to identify adverse events that occur at significantly higher rates on the drug of the interest than on comparators. Usually the data from all trials are combined in a meta-analysis. This method will identify most common adverse reactions but it has important limitations, as follows:
Numbers of patients studied (on average 1,500 – not enough to identify rare ADRs)
Short duration of follow-up (weeks or months rather than years)
Selection of patients (those at greatest risk of ADRs may be excluded)
Artificial conditions (e.g. close monitoring).
Sometimes clinical trials are conducted after marketing. These may provide important safety information provided that they contain enough patients, and have few exclusion criteria and outcomes that are easily measured (e.g. mortality). This type of study is known as the large simple trial
Post-marketing safety monitoring
Because of the limitations of pre-marketing studies described above, safety can only be regarded as provisional when a new medicine is first marketed and there is a need to collect more evidence arising from "real world" usage. The principal mechanisms for monitoring the safety of medicines in the UK are the Yellow Card Scheme and Prescription-Event Monitoring. Both methods depend on the co-operation of healthcare professionals, and traditionally have received excellent support on a voluntary basis. They are described in more detail below.
Yellow card reporting system
The idea of reporting ADRs came from the thalidomide tragedy in the early 1960s. There were few regulatory controls at that time and thalidomide had not been through the kind of testing programme described above before it was marketed and used frequently by pregnant women. This drug is a teratogen and if used in early pregnancy causes major limb deformities known as phocomelia.
Picture of Thalidomide Phocomelia
Phocomelia is very rare in the absence of thalidomide exposure but about 3 years after the drug was first marketed, an epidemic occurred in Germany. Initially the cause was thought to be environmental contamination but eventually the link with thalidomide was suggested by a doctor who had seen 3 cases. The drug was soon withdrawn and it became apparent that, worldwide, about 10,000 fetuses had been affected.
The UK reporting scheme which was developed as a consequence of thalidomide is known as the Yellow Card Scheme. It is run by the Medicines and Healthcare Products Regulatory Agency (MHRA). About 20,000 reports are submitted each year by health professionals including doctors, dentists, pharmacists and nurses. Since the scheme started in 1964 many important new adverse reactions have been identified by this "early warning" system. Reporting adverse reactions experienced by your patient helps to prevent such reactions occurring in others. The GMC considers this to be one of the professional duties of a doctor.
The purpose of yellow cards is simply to provide a mechanism whereby the association between drug and an unexpected reaction can be identified as quickly as possible. It relies on doctors (and other health professionals such as pharmacists) voluntarily reporting their suspicions. These are collated centrally and carefully screened for unrecognised safety hazards. A yellow card is simple to complete and postage is pre-paid. Reports are acknowledged and information is made available from the database on request. Evaluated data from the scheme are also published in the bulletin Drug Safety Update which is available on the MHRA's Website.
Yellow cards are readily available – there are several in the back of the BNF and an electronic version is now available here. The scheme applies to all medicines and any ADR may be reported. It is most important to fill in a yellow card when:
The drug suspected is new (as indicated by the inverted black triangle symbol – ▼ – this is included on product information and in the BNF)
The reaction is serious or has a fatal outcome
The reaction is not recognised in the product information or BNF.
Don’t be off by uncertainty – reasonable suspicion is worth reporting. If in doubt, it’s safer to report. All information provided is treated in strict confidence.
Prior to 2005, only health professionals were allowed to submit yellow card reports but since then patients and carers have also been allowed to report. The value of patient reporting is still being assessed.
Internationally, most developed countries have ADR reporting schemes and worldwide data are held by the World Health Organisation at a monitoring centre in Sweden.
Prescription event monitoring
Prescription Event Monitoring (PEM) is a system whereby patients taking specific medicines are identified through GP prescriptions. Events which occur during the subsequent 6 months are captured on green forms which are sent to and completed by the GP. The scheme covers the whole of England and has been operated since 1980 by the Drug Safety Research Unit (an independent charitable trust) in Southampton. To date, around 80 drugs have been studied by this method. Similar schemes have also been developed in New Zealand and Japan.
PEM is mostly focussed on new medicines, particularly those used for chronic diseases, and is complementary to yellow cards as a method of identifying unexpected ADRs. It has the advantage the number of users is known and therefore event frequencies can be accurately quantitated. An important difference between PEM and yellow cards is that all events are requested, whether or not there is suspicion that it was drug-induced. The method may therefore identify effects that clinicians do not recognise as being ADRs.
The usual size of the cohort in PEM is about 10,000 patients – almost an order of magnitude greater than the usual number studied in clinical trials. Thus, when a medicine has been studied by PEM and no important new ADRs have been identified, the data provide considerable reassurance about its safety. However, PEM studies are not large enough to identify very rare ADRs (which can only be picked up through yellow cards). The PEM scheme is voluntary and receives excellent co-operation from GPs, with response rates of around 60%. Feedback is provided through a newsletter and the study findings are published in the literature.
PEM is a well-established part of the safety net. Completing green forms is simple, helps to protect patients from ADRs and is considered to be a professional duty by the UK General Medical Council. No judgement is required as to when to complete a green form – all forms should be returned, even if there are no events to report. As with yellow cards, all information provided is treated in strict confidence.