A Textbook of Clinical Pharmacology and Therapeutics

(nextflipdebug2) #1

64 ADVERSE DRUG REACTIONS


system can offer the 100% accuracy that current public opinion
expects. The ideal method would identify adverse drug reactions
with a high degree of sensitivity and specificity and respond
rapidly. It would detect rare but severe adverse drug reactions,
but would not be overwhelmed by common ones, the incidence
of which it would quantify together with predisposing factors.
Continued surveillance is mandatory after a new drug has
been marketed, as it is inevitable that the preliminary testing
of medicines in humans during drug development, although
excluding many ill effects, cannot identify uncommon adverse
effects. A variety of early detection systems have been intro-
duced to identify adverse drug reactions as swiftly as possible.


PHASE I/II/III TRIALS

Early (phase I/II) trials (Chapter 15) are important for assess-
ing the tolerability and dose–response relationship of new
therapeutic agents. However, these studies are, by design,
very insensitive at detecting adverse reactions because they
are performed on relatively few subjects (perhaps 200–300).
This is illustrated by the failure to detect the serious toxicity
of several drugs (e.g. benoxaprofen,cerivastatin,felbamate,
dexfenfluramineandfenfluramine,rofecoxib,temofloxacin,
troglitazone) before marketing. However, phase III clinical
trials can establish the incidence of common adverse reactions
and relate this to therapeutic benefit. Analysis of the reasons
given for dropping out of phase III trials is particularly valu-
able in establishing whether common events, such as
headache, constipation, lethargy or male sexual dysfunction
are truly drug related. The Medical Research Council Mild
Hypertension Study unexpectedly identified impotence as
more commonly associated with thiazide diuretics than with
placebo or β-adrenoceptor antagonist therapy. Table 12.3
illustrates how difficult it is to detect adverse drug reactions
with 95% confidence, even when there is no background inci-
dence and the diagnostic accuracy is 100%. This ‘easiest-case’
scenario approximates to the actual situation with thalido-
mide teratogenicity: spontaneous phocomelia is almost
unknown, and the condition is almost unmistakable. It is
sobering to consider that an estimated 10 000 malformed
babies were born world-wide before thalidomidewas with-
drawn. Regulatory authorities may act after three or more
documented events.
The problem of adverse drug reaction recognition is much
greater if the reaction resembles spontaneous disease in the
population, such that physicians are unlikely to attribute the
reaction to drug exposure: the numbers of patients that must
then be exposed to enable such reactions to be detected are
greater than those quoted in Table 12.3, probably by several
orders of magnitude.


YELLOW CARD SCHEME AND POST-MARKETING
(PHASE IV) SURVEILLANCE

Untoward effects that have not been detected in clinical trials
become apparent when the drug is used on a wider scale. Case


reports, which may stimulate further reports, remain the most
sensitive means of detecting rare but serious and unusual
adverse effects. In the UK, a Register of Adverse Reactions
was started in 1964. Currently, the Medicines and Healthcare
products Regulatory Agency (MHRA) operates a system of
spontaneous reporting on prepaid yellow postcards. Doctors,
dentists, pharmacists, nurse practitioners and (most recently)
patients are encouraged to report adverse events whether actu-
ally or potentially causally drug-related. Analogous schemes are
employed in other countries. The yellow card scheme consists
of three stages:


  1. data collection;
    2.analysis;
    3.feedback.
    Such surveillance methods are useful, but under-reporting is a
    major limitation. Probably fewer than 10% of appropriate
    adverse reactions are reported. This may be due partly to con-
    fusion about what events to report, partly to difficulty in rec-
    ognizing the possible relationship of a drug to an adverse
    event – especially when the patient has been taking several
    drugs, and partly to ignorance or laziness on the part of poten-
    tial reporters. A further problem is that, as explained above, if
    a drug increases the incidence of a common disorder (e.g.
    ischaemic heart disease), the change in incidence must be very
    large to be detectable. This is compounded when there is a
    delay between starting the drug and occurrence of the event
    (e.g. cardiovascular thrombotic events including myocardial
    infarction following initiation of rofecoxibtherapy). Doctors
    are inefficient at detecting such adverse reactions to drugs,
    and those reactions that are reported are in general the obvi-
    ous or previously described and well-known ones. Initiatives
    are in progress to attempt to improve this situation by involve-
    ment of trained clinical pharmacologists and pharmacists in
    and outside hospitals.
    The Committee on Safety of Medicines (CSM), now part of
    MHRA, introduced a system of high vigilance for newly mar-
    keted drugs. For its first two years on the general market, any
    newly marketed drug has a black triangle on its data sheet and
    against its entry in the British National Formulary. This con-
    veys to prescribers that any unexpected event should be
    reported by the yellow card system. The pharmaceutical com-
    pany is also responsible for obtaining accurate reports on all
    patients treated up to an agreed number. This scheme was
    successful in the case of benoxaprofen, an anti-inflammatory


Table 12.3:Numbers of subjects that would need to be exposed in order to
detect adverse drug reactions

Expected frequency Approximate number of patients
of the adverse effect required to be exposed

For one event For three events

1 in 100 300 650
1 in 1000 3000 6500
1 in 10 000 30 000 65 000
Free download pdf