Intended for healthcare professionals


Monitoring drug treatment

BMJ 2003; 327 doi: (Published 20 November 2003) Cite this as: BMJ 2003;327:1179
  1. Munir Pirmohamed, professor of clinical pharmacology (munirp{at},
  2. Robin E Ferner, clinical pharmacologist (R.E.Ferner{at}
  1. Department of Pharmacology and Therapeutics, University of Liverpool, Liverpool L69 3GE
  2. West Midlands Centre for Adverse Drug Reaction Reporting, City Hospital, Birmingham B18 7QH

    Criteria used for screening tests should apply to monitoring

    We all want our treatments to work, and none of us wishes treatment to cause harm. Monitoring drug treatment is one way of seeing that a treatment works, while protecting the patient from adverse drug effects. For many patients and many treatments clinical evaluation is sufficient. An example is measuring the blood pressure in a patient on antihypertensive treatment. When therapeutic goals cannot always be directly observed, monitoring may require blood tests in order to know whether they have been reached. An obvious example is the measurement of the international normalised ratio (INR) in patients treated with warfarin. As well as ensuring that the therapeutic goal, the prevention of thrombosis, is likely to be met,1 measuring the INR helps to avoid the risk of haemorrhage, which rises steeply as the INR increases above 2.0.2

    Monitoring treatment to anticipate or detect adverse reactions to drugs before they become inevitable or irreversible is clearly important. Upwards of half of the entries in the electronic Medicines Compendium (eMC) suggest monitoring of one kind or another.3 However, for a monitoring test for an adverse drug reaction to be useful clinically, it should satisfy much the same criteria as have been put forward for screening tests (box).4 When the criteria are met, monitoring can be very effective. This is exemplified by the successful use of monitoring in patients treated with clozapine, an atypical antipsychotic, associated with agranulocytosis in 0.8% of patients.5 All patients who are taking clozapine have white cell counts performed weekly for the first 18 weeks of treatment and less often thereafter.6 Clear criteria exist for when the drug should be withdrawn, and patients continue treatment only if the white cell count is satisfactory. This strategy has reduced the incidence of clozapine induced agranulocytosis and prevented deaths from a serious adverse reaction.7 The scheme's success is largely the result of frequent monitoring at the time when the risk of agranulocytosis is highest and of the clear rules for action if results are abnormal. The adverse reaction evolves slowly enough for once weekly monitoring to be effective. By contrast serious hyperkalaemia could occur at any time in patients treated for heart failure with spironolactone plus an angiotensin converting enzyme inhibitor and evolve rapidly to cause lethal arrhythmia. Thus annual measurement would be of little help in avoiding serious effects.8

    Factors to take into account when monitoring for an adverse drug effect (adapted from reference 4)

    The adverse effect

    The effect should be potentially serious

    The relation between the latent and overt effects should be known

    The monitoring test

    The test should be safe, simple, precise, and validated

    The distribution of test values in the exposed population should be known and suitable cut-off values established

    The test should be acceptable to treated patients

    A strategy in the face of a positive monitoring test should be agreed

    The response to positive tests

    An effective intervention should exist

    This early intervention should make the outcome better than it would have been with delayed intervention

    Evidence for the intervention should be robust

    The monitoring strategy

    The strategy should reduce morbidity or mortality from the adverse effect

    The strategy should be acceptable to patients and professionals

    Benefits of monitoring should outweigh the physical and psychological harm

    The cost of monitoring should be proportionate

    A system for assuring the standards of the monitoring programme should exist

    Possibility of reducing or removing risks of adverse effects by selection of drug or dosage, or by pretreatment detection of susceptible people, should have been fully explored

    An area where monitoring is commonly recommended, but where the criteria are difficult to satisfy, is in the detection of drug induced liver injury. Statins, for example, can increase serum activity of transaminase in about 3% of patients and rarely can lead to symptomatic hepatic damage.9 This has prompted recommendations for monitoring in the product literature; however, guidelines for different statins differ both in recommended frequency of monitoring and advice on the action to take if hepatic abnormalities are detected. The problem is compounded by the fact that, firstly, we do not understand the relationship between mild abnormalities of liver function and symptomatic liver injury, since liver function may improve even with continued treatment with statin10; secondly, it is unclear if or when treatment should be stopped; and, thirdly, infrequent monitoring as currently recommended is likely to miss most patients who develop the sudden idiosyncratic hepatic reactions. Monitoring for liver damage from statins may anyway be unnecessary—a meta-analysis examining 112 000 person years of exposure to pravastatin found the frequency of abnormal liver function tests (1.4%) to be similar in statin and placebo arms,11 and in the heart protection study treatment with statins at high dose (40 mg simvastatin) seemed safe.12 When considered with evidence about muscle damage from statins10 the findings imply that these drugs can be used without any regular monitoring. This was the conclusion of a retrospective analysis of 1014 patients in primary care, where the occasional finding of abnormal laboratory values rarely resulted in drug discontinuation.13 A policy of non-monitoring would prevent the unnecessary discontinuation of statins and possibly permit these drugs to become available over the counter in the future.14

    Product information on drugs often suggests monitoring of one kind or another but does not specify the frequency of testing or the strategy to adopt if tests are positive, and many of the proposed tests fail to satisfy the criteria we have listed. We need better evidence on which to base our monitoring strategies. Meanwhile, adverse reactions will often be prevented more effectively (and economically) by educating prescribers and increasing patients' awareness than by empirical blood test monitoring. After all, rational therapeutics demands a more careful approach to drug treatment than simple opportunistic measurement in the outpatient clinic.

    See also 1222


    • Competing interests MP is a member of the Committee on Safety of Medicines (CSM), and MP and REF are members of the CSM Subcommittee on Pharmacovigilance, but the views expressed here are their own. MP has current grant support from Astra-Zeneca, Pfizer, and Bristol Myers Squibb for research into adverse drug reactions.


    View Abstract