Pharmaceutical Journal Vol 263 No 7059 p285-288
August 21, 1999 Articles

Pharmacovigilance resources in the United Kingdom

By Ian C. K. Wong, PhD, MRPharmS

Dr Wong is a lecturer in pharmacy practice at the Pharmacy Practice Research Unit, School of Pharmacy, University of Bradford, Bradford, BD7 1 DP (tel 01274 234764; fax 01274 234769; e–mail: i.c.k.wong@bradford.ac.uk)

This article provides readers with a brief introduction to the main pharmacovigilance resources in the United Kingdom.
The article explains how the various systems work, focuses on their strengths and limitations, and looks to their future

Since April, 1997, all United Kingdom hospital pharmacists have been eligible to report adverse drug reactions (ADRs) to the Medicines Control Agency (MCA) and the Committee on Safety of Medicines (CSM) via the yellow card scheme. This effort has already made a valid contribution to the process of pharmacovigilance.¹
Pharmacovigilance can be defined as a process of identifying and responding to risk-benefit issues arising with marketed medicines.² This process requires input from various specialties such as clinical pharmacology and epidemiology. Therefore, it is not too surprising that most of the techniques and study designs employed in pharmacovigilance are those used in epidemiology. When a suspected ADR is assessed, the investigator may apply the principles of clinical pharmacology to investigate the effects of the drug on an individual patient. Many ADR algorithms and tables have been developed to assist the investigator with causality assessment. Pharmacists are familiar with these algorithms and tables, as they are taught in the undergraduate and post-registration training.³ On the other hand, only epidemiological techniques can quantify the risks of such ADRs. Large databases are essential for conducting large epidemiological studies, especially for the study of rare ADRs. For example, it is necessary to collect data from 6,000 patients in order to detect an ADR with an incidence of 1 in 2,000.⁴
Some pharmacists are not aware of the details of some of these powerful databases which are available in the UK.⁵ The purpose of this article is to introduce pharmacists to these resources.

Spontaneous reporting systems — yellow card scheme

The yellow card scheme was established in 1964 as a result of the thalidomide tragedy. Since then, the system has become one of the major international pharmacovigilance resources. The yellow card scheme is run jointly by the MCA (the regulatory agency) and the CSM (an expert advisory committee to the MCA).
Since 1991, the yellow card scheme has been enhanced by a new computer system, the ADROIT (Adverse Drug Reaction On–line Information Tracking) system⁶. ADROIT is different from other databases. Not only does it store the details of the report, but also the image of the yellow card in the optical system. Multiple users can view any yellow card on screen at the same time.
The MCA receives approximately 20,000 yellow cards each year. The reports are prioritised so that serious adverse drug reactions receive early attention. The yellow cards are classified into seven priorities by a member of the scientific staff according to the drugs and the nature of the ADR. Priority 1 reports receive the earliest attention.
The details of each yellow card are entered into ADROIT by an assistant scientific classifier. There are a number of quality assurance steps, where the details are checked by more senior scientific staff such as scientific assessors or medical assessors, before the report is committed to the database. This process ensures a high quality of data.⁷

Adverse drug reaction signal generation All serious ADRs to established drugs are reviewed by a group of medical and scientific staff in a weekly drug safety review. Safety issues raised at the meetings are further investigated. ADRs to drugs under intensive monitoring (“black triangle” drugs) are monitored by individual assessors and reviewed at monthly meetings. Drug safety issues highlighted in these two systems can be further reviewed by the pharmacovigilance subcommittee of the CSM. The whole process is summarised in Figure 1.

Figure 1. The Adverse Drug Reaction On-line Information Tracking (ADROIT) and yellow card system

• The product licence for the drug may be withdrawn by the regulatory authority, eg, as with triazolam⁷
• The company might voluntarily suspend or withdraw the product (recent examples are sertindole and tolcapone⁸)
• If only a particular group of patients is at risk, the product licence may be amended to restrict or prevent these patients from being exposed to the risks (for example, terfenadine has recently been reverted from a pharmacy medicine to prescription only medicine⁹)

The summary of product characteristics and the patient information leaflet is updated and the information disseminated to health professionals via the bulletin Current Problems in Pharmacovigilance or a “Dear Doctor/Pharmacist” letter or both.
Strengths of the scheme The yellow card scheme is useful for detecting very rare adverse reactions. It is relatively inexpensive when compared with other postmarketing surveillance tools. It is also useful for generating “signals”, ie, hypotheses that a particular drug may cause a particular ADR. The scheme allows for continual safety monitoring of a product within its life span. Safety problems may not arise until the product has been marketed long–term, for example, as with high-strength lipase pancreatins.
When combined with drug usage data, the scheme allows for comparisons of adverse reaction “profiles” between products within the same therapeutic or pharmacological class, for example, gastrointestinal ulcers caused by non-steroidal anti-inflammatory drugs (NSAIDs).
In some cases, the scheme can provide information on factors which predispose patients to ADRs. However, it relies on the completeness of the yellow card.
Limitations of the scheme One limitation of the scheme is under-reporting. Also, the scheme requires health professionals to recognise ADRs which may be affected by the fact that most of the ADRs mimic naturally occurring illnesses.
The scheme is unable to provide incidence rates because of a lack of denominator data and the reporting rate is affected by publicity.
Future of the scheme and ADROIT The yellow card scheme will remain the main source of pharmacovigilance data in the UK as it has input from most UK health professionals. Furthermore, ADROIT is a powerful computer system, and it will be widely used as a model by other European countries.

Prescription events monitoring systems

Prescription event monitoring (PEM) is a method pioneered by Professor William Inman of the Drug Safety Research Unit in Southampton. PEM aims to complement the yellow card scheme. It monitors drug safety by direct contact with the patient’s own general practitioner. All prescriptions that are written in England pass through the Prescription Pricing Authority, which is able to extract data for drugs of special interest and send these data to the Drug Safety Research Unit. After a particular period (normally six months), the Drug Safety Research Unit sends special questionnaires (green forms) to general practitioners requesting information about any “event” which was considered to be of sufficient importance to enter into the patient’s notes, since the drug was first prescribed, and demographic details such as date of birth and sex10. Figure 2 summarises the prescription event monitoring process.

Adverse drug reaction signal generation Most ADRs occur during the early period of treatment, especially during the first month. Therefore, if any particular event shows an excessively high rate during the first month compared with that during months 2 to 6 after the treatment started, this is suggestive of the event possibly having been an ADR. Since more than 50 PEM studies have been completed, the Drug Safety Research Unit is able to calculate the average rate of each event. If the event rate for a particular drug is higher than the average rate, it signals a possible ADR.¹¹ Finally, reasons for discontinuation can also be used to signal a possible adverse drug reaction. For example, the top reason for patients discontinuing lansoprazole during the first month of treatment was diarrhoea (2.1 per cent of the cohort). Diarrhoea was also the highest reported event during the first month of treatment.¹²
Strengths of prescription event monitoring PEM actively encourages prescribers to report adverse events. Around 50 to 70 per cent of GPs will respond to the questionnaire (green form), which is different from the yellow card. PEM is non-interventional and represents real-life clinical use. It allows surveillance of deaths and pregnancies.
PEM is relatively inexpensive to set up. If long latency ADRs are suspected, a follow up questionnaire can be sent and long-term safety data can be obtained.¹³
Limitations of prescription event monitoring Any drug included in PEM must be prescribed by general practitioners on a scale sufficient to allow an adequate group of patients to be assembled within a reasonable time. Therefore, drugs which are mainly used in hospitals are not suitable for this kind of study.
As the participation of general practitioners in PEM is voluntary and without financial incentive, at least 30 per cent of GPs choose not to complete the green forms. This creates a possible bias and the effect of these non–responders on the PEM study is still unknown. PEM has limited capability because it needs data from more than 10,000 questionnaires for each drug before the full analysis. This is highly labour-intensive and is not a cost-effective process.
No accurate estimate of compliance (whether or not the patient takes the medication) is known.
Future of prescription event monitoring PEM has been described as the biggest advance in pharmacovigilance since the development of the yellow card scheme. The Select Committee on the European Communities recommended that the European Commission should consider the use of PEM throughout the Community. However, uncertainties about funding limits the future development of PEM.
Furthermore, as PEM takes a long time to gather information from the PPA and GPs, it is unable to respond to urgent drug safety issues. Unless PEM data have already been collected, the system is less attractive than the automated databases described below.

General Practice Research Database (formerly VAMP database)

Computerisation in general practice revolutionised patient record management. In the late 1980s, VAMP Health, a commercial company, designed and marketed a GP computer system which allowed for comprehensive recording of clinical and demographic information for individual patients on personal computers. The information recorded in this computer system is highly valuable for use in pharmacoepidemiology and pharmacovigilance.
GPs were trained to enter clinical data items in a standard manner. These clinical data were then provided to VAMP in an anonymised form, thus fully preserving patient confidentiality. Furthermore, GPs also provided photocopies of referral letters to VAMP in order to validate the information in the database. In return, GPs received financial compensation for their co-operation.
As in other pharmacoepidemiological studies, a high quality of data is of the utmost importance. In order to validate the data, VAMP entered a co-operative agreement with the Boston Collaborative Drug Surveillance Program, which conducted extensive studies involving thousands of patients. In these studies, the diagnoses presented on photocopies of referral letters from hospital consultants were compared with the diagnoses recorded on the computer. The concordance was greater than 90 per cent. The whole process is summarised in Figure 3.
Since 1994, the VAMP database has been officially transferred to the Office of National Statistics to be managed on behalf of the Department of Health and was renamed the General Practice Research Database (GPRD). Currently, around 1,500 general practitioners with a population coverage in excess of three million people are included in the GPRD.
Strengths of the General Practice Research Database One advantage of the GPRD is that drug safety studies can be carried out rapidly in response to the urgent drug safety inquiries. For example, a study on oral contraceptives and venous thromboembolism was completed within six months.¹⁴
Flexibility in the methodology, in both cohort and case control studies, is possible. Control groups can be identified and used for comparison. Furthermore, studies can be either disease-orientated or drug-orientated.
Both denominators (number of patients exposed to drugs) and numerators (number of patients experiencing the ADRs) are available; therefore, reliable incidence rates can be calculated.
Patients’ lifetime medical records are available; therefore some supplementary data are available when required.

Limitations of the General Practice Research Database A limitation of the GPRD is that it is difficult to study drugs used mainly in the hospital setting, eg, chemotherapeutic agents.
GPs do not routinely record social data (eg, occupation, employment status, marital status) about their patients in a standardised format, thus, although such details may be recorded, their accuracy in terms of both description/categorisation and current validity is likely to be variable.
Most of the studies have been done retrospectively; therefore some data may not be available.

MEMO — Tayside Medicines Monitoring Unit

The Medicines Monitoring Unit (MEMO) is an independent organisation based at the University of Dundee. The MEMO has access to records of 400,000 patients registered with GPs in Tayside, where each patient has a unique community health index number (CHNo). This CHNo allows the MEMO to link the data from different sources, such as dispensed prescriptions, hospital morbidity data, and death certificates, as well as certain types of medical procedures. This system is illustrated in Figure 4. For example, the MEMO has conducted a cohort study of 52,293 NSAID users as a means of ranking the order of toxicity obtained for different NSAIDs.¹⁵ The study showed good concordance with other studies.
Strengths and limitations of the Medicines Monitoring Unit The MEMO has similar strengths and weaknesses as the GPRD. It has access to hospital records, whereas the GPRD has access to GP records. Furthermore, since the drug utilisation data are derived from dispensed prescriptions, it eliminates primary patient non-compliance (ie, when patients do not even obtain their medication).
The major weakness of the MEMO is that the database is relatively small and can only be used to study widely used drugs, such as NSAIDs and antibiotics. Furthermore, prescription data are captured manually at the moment.
Future of the Medicines Monitoring Unit The MEMO is aiming to extend the record-linkage studies to the whole of Scotland. This will increase the covered population to 5.5 million, and the feasibility of such studies is currently being tested. If the MEMO successfully extends the studies, it will certainly become a very powerful pharmacoepidemiological tool.
Furthermore, the MEMO is investigating the feasibility of capturing prescription data directly from individual pharmacies via electronic communication. It also gives an added advantage if pharmacies enter data for over-the-counter products. The MEMO would be the only database in the UK with this capability.

Summary

Information technology has already revolutionised pharmacovigilance by introducing ADROIT, the GPRD and the MEMO, and there is no doubt that these databases will continue to develop and provide better information for the regulatory authorities, manufacturers and health professionals on the risks and benefits of medical interventions.
None of the pharmacovigilance methods is ideal; however, they all complement each other and working together will achieve the goal of safe use of medicines.

References