Principles and Practice of Pharmaceutical Medicine

the physician decided to stop treatment. Now we
see that the one patient in five who is ‘holiday-
prone’ is abruptly stopping treatment on a more or
less monthly basis!
Urquhart has pointed out that pharmacological
evaluation of new drugs is strongly biased toward
studies of the onset of drug action as dosing com-
mences, ignoring the offset of drug action as dosing
stops. As we now see the frequency with which
sudden stops in dosing occur, it is time for pharma-
cologists to rebalance their scientific focus on
the stopping as well as the starting of drug dosing
and their associated pharmacodynamical conse-
quences.
Another consideration is drugs that have so-
called ‘first-dose’ effects: dosing cannot begin at
the full therapeutic level, but must be gradually
increased from a low, initial dose. Some of these
agents may be especially hazardous in the holiday-
prone patient, as longer holidays permit the patient
to return far enough toward the drug-naive state,
with overdose toxicities developing in the wake of
sudden resumption of full-strength dosing as the
drug holiday ends.

Statistics

Numerous studies show that methods which afford
patients easy ability to censor evidence for poor
compliance consistently result in gross overesti-
mates of compliance (Waterhouseet al., 1993).
This finding poses a major challenge to contem-
porary clinical trials design and analysis. The pre-
vailing statistical policy of ITTanalysis is to ignore
all information on actual drug exposure and simply
average drug responses in all patients randomized
to receive drug, irrespective of how much drug they
took, or whether they took any at all. ITTanalysis is
far more seriously biased than previously believed
by the pharmacodynamic skewing arising from,
not only simple underdosing, but erratic patterns
of dosing likely to trigger recurrent rebound effects
or first-dose effects. If clinical trialists have been
slow to grasp the implications of these dosing
patterns, the problem has certainly attracted the
attention of the biostatistical community, leading
to a series of symposia throughout the 1990s and

into the next decade. Statistical research done

under the rubric of ‘causal inference’ is at the

forefront of the effort to integrate drug exposure

information and its clinical correlates into clinical

drug development.

Diversity of electronically monitored

packages

Several firms are beginning to provide electroni-

cally monitored packages: vials, blisters and nebu-

lizers.

Pharmaceutical industry

Several major pharmaceutical firms have begun to

address the issue defined by Norell, which might be

paraphrased as answering the question ‘how much

compliance is enough?’ (Urquhart, 1993). The

answer to this question can support comparative

claims for superiority in maintenance of therapeu-

tic action in the face of common lapses in compli-

ance. This is a therapeutically sound and

potentially important new area of comparative

pharmaceutical advantage. It can be expected to

grow as a marketing issue, driven by the increasing

orientation toward the outcomes of pharmaceutical

care.

Packaging industry

For the packaging industry, the incorporation

of compliance monitoring into packaging has

value-added potential, the realization of which

depends on many factors. The key factor, naturally,

is the definition of cost-effectiveness of providing

compliance information in specific therapeutic

areas.

As usual with an emerging medical capability,

there is an element of ‘chicken–egg’ impasse that

has to open up in order for volume to increase, costs

of goods to fall and perceived value of the informa-

tion to grow. Obviously, high prices can hinder

growth and diminish the perceived value of the

information, but pricing linked to cost-of-goods

27.14 WHAT ARE THE FACTS WHICH DEMONSTRATE THE IMPORTANCE OF COMPLIANCE? 367