Intended for healthcare professionals


A meta-analysis of the association between adherence to drug therapy and mortality

BMJ 2006; 333 doi: (Published 29 June 2006) Cite this as: BMJ 2006;333:15

A different view of the 'effect' of adherence to dosing regimens for trial placebos


The meta-analysis by Simpson et al. (1) shows a strong inverse
association between adherence with placebo regimens and mortality in 21
trials. The conclusion reached by the review, and the associated
commentary by Chewning (2), was that the explanation is to be found in a
presumed (but to my knowledge never really studied with reliable methods)
association between adherence with drug dosing regimens and with various
healthy lifestyle factors, e.g., exercise, diets high in fruits &
vegetables, non-smoking.

Almost all of these trials involved the use of non-trial medications
(and maybe all, for it is unclear from the papers what the extent of use
of non-trial medicines actually was). Tight coupling between compliance
with trial agents (either active or placebo) and compliance with non-trial
medicines could create the appearance of an ‘effect’ of variable
compliance with the trial's placebo. A non-trial medicine, by definition,
is one deemed too important to the patient's care to be discontinued for
the purposes of improving trial sensitivity. Poor adherence to the
prescribed dosing regimens for non-trial medicines can thus reasonably be
expected to be harmful, the more so with more severe disease.
Electronically compiled dosing histories of concomitantly prescribed
medicines support the conclusion of a strong association between adherence
with one and all co-prescribed medicines (3), but more robust analysis of
this association is needed.

Neither of the first two big primary prevention trials of lipid-
modifying agents found any relation between adherence with placebo and
lipid changes (4). These were studies of patients who, at enrolment, had
elevated lipid levels but were free of overt disease. Though some were,
in the course of these multiyear trials, to contract overt coronary heart
disease and/or other conditions, for which non-trial medicines would have
been prescribed, the prevalence of non-trial medicines use would have been
low. This difference is reflected in the 8.4-fold higher mortality in the
Coronary Drug Project Trial (CDPT) than in the aforementioned primary
prevention trials (4). All patients enrolled into CDPT had had a prior
myocardial infarction, ~half of whom at the start of the trial were taking
non-trial medicines indicative of a diagnosis of congestive heart failure
(4). Alas, no attempt was made to capture evolving changes in prescribed
non-trial medicines in the course of that study.

Today, it is clear that evolving histories of drug
prescribing/dispensing are, when complete, an economical source of good
information on the evolution of each patient’s diseases and their
respective severity -- a cornerstone of the “Dutch school” of
pharmacoepidemiology in the past 2 decades and a useful window onto
changing health status of patients over time (5). Also, today
electronically compiled drug dosing histories can provide much more
reliable and complete information on patients’ actual exposure to
prescribed drugs than was possible when most of the studies in the meta-
analysis were run (6,7). Thus, the usual conclusion is ‘more work has to
be done’, especially on the links between adherence to (a) trial agents,
(b) non-trial drugs, (c) lifestyle factors, before leaping to one view or
the other. Perhaps both (b) and (c) play a role, but how strong is each?

1. Simpson SH, Eurich DT, Majumdar SR, Padwal R, Tsuyuki RT, Varney J,
Johnson JA. A meta-analysis of the association between adherence to drug
therapy and mortality, BMJ 2006; 333:15-20.
2. Chewning B. The healthy adherer and the placebo effect. BMJ
3.. Cramer J, Vachon L, Desforges C, Sussman NM. Dose frequency and dose
interval compliance with multiple antiepileptic medications during a
controlled clinical trial. Epilepsia 1995; 36: 1111-7.
4. Urquhart J. Patient compliance as an explanatory variable in four
selected cardiovascular trials. In: Patient Compliance in Medical
Practice and Clinical Trials. Eds Cramer JA, Spilker B. New York: Raven
Press, 1991, 301-322.
5. Leufkens HG, Urquhart J. Automated record linkage in The Netherlands.
Chapter 20 in: Pharmacoepidemiology, 4th Ed. Strom B., ed. Chichester
(UK): John Wiley, 2005, 311-22.
6, Liu H, Golin CE, Miller LG, Hays RD, Beck CK, Sanandaij S, Christian J,
Maldonado T, Duran D, Kaplan A, Wenger NS. A comparison study of multiple
measures of adherence to HIV protease inhibitors. Ann Intern Med 2001;
134: 968-77.
7. Vrijens B, Tousset E, Rode R, Bertz R, Mayer S, Urquhart J. Successful
projection of the time-course of drug concentration in plasma during a one
-year period from electronically compiled dosing-time data used as input
to individually parameterized pharmacokinetic models. J Clin Pharmacol
2005; 45: 461-7.

Competing interests:
I am a partner in AARDEX Ltd, pioneer developer of electronic means for compiling drug dosing histories in ambulatory patients.

Competing interests: No competing interests

08 July 2006
John Urquhart, MD, FRCPE, FRSE
Chief Scientist, AARDEX Ltd; Emeritus Professor of Pharmacoepidemiology, Maastricht University;
975 Hamilton Ave, Palo Alto, CA 94301 USA