Drug and Alcohol Dependence 54 (1999) 171–177
‘Research’ versus ‘real-world’ patients: representativeness of
participants in clinical trials of treatments for cocaine dependence
Kathleen M. Carroll
a,
*, Charla Nich
a
, A. Thomas McLellan
b
, James R. McKay
b
,
Bruce J. Rounsaville
a
a
Di6ision of Substance Abuse, Department of Psychiatry, Yale Uni6ersity School of Medicine, New Ha6en CT
06519
, USA
b
Department of Psychiatry, Uni6ersity of Pennsyl6ania, Pennsyl6ania PA
19104
, USA
Received 13 May 1998; accepted 16 September 1998
Abstract
Rigorous clinical trials have been criticized as having limited external validity, and specifically that subjects participating in
clinical trials are not representative of individuals seen in clinical practice. To assess the representativeness of subjects participating
in clinical trials, 243 research subjects participating in clinical trials of treatments for cocaine dependence were compared to a
sub-sample of 213 individuals being treated for cocaine dependence in outpatient clinical settings from a large national database.
The data suggest that research findings are not invariably based on less challenging patients with mild forms of substance
dependence and related problems; moreover research patients may be similar to, if not more severe than, individuals with cocaine
problems seen in regular clinical settings in the community. © 1999 Published by Elsevier Science Ireland Ltd. All rights reserved.
Keywords
:
Cocaine treatment; Generalizability; Clinical trials
1. Introduction
The level of methodological rigor associated with the
conduct of clinical trials of treatments for substance use
disorders has risen dramatically in recent years.
Methodological features such as random assignment,
standardized diagnostic criteria to define the study sam-
ple, double-blind procedures, standardized assessments
with established psychometric properties, biological
measures of outcomes including urine toxicology
screens, fidelity checks to assure that treatments have
been implemented consistently and as defined in the
treatment protocol, and appropriate and often multiple
control conditions have all become virtual requirements
in research of this type. The controls, standards, and
constant monitoring that are now part of ‘rigorous’
clinical efficacy research have been seen as scientific
advances as they allow new levels of inference regarding
the mechanisms by which treatments work, as well as
the ability to infer causality regarding patient outcomes
following treatment (Chambless and Hollon, 1998).
A major issue raised by the heightened standard of
efficacy research is the degree to which findings gener-
ated by this type of research are applicable in regular,
‘real world’ clinical settings. Rigorous clinical trials are
frequently criticized in this regard (Institute of
Medicine, 1998; Seligman, 1995; Russell and Orlinsky,
1996; Peele, 1998). For example, Weisz et al. (1995)
suggested that many subjects in clinical trials of anxiety
disorders in children were ‘not representative of those
in current clinical practice’. In contrast, a number of
recent findings suggest that methodological features
designed to protect internal validity are not necessarily
incompatible with external validity. For example,
McKay and colleagues found that subjects who refused
randomization in a clinical trial evaluating outpatient
versus day hospital treatment for alcohol dependence
had somewhat different admission profiles but com-
* Corresponding author. Tel.: + 1 203 7897080 ext. 336.
0376-8716/99/$ - see front matter © 1999 Published by Elsevier Science Ireland Ltd. All rights reserved.
PII: S0376-8716(98)00161-6