Letters to the Editor
RE: ‘‘QUALITY OF REPORTING OF OBSERVATIONAL LONGITUDINAL RESEARCH’’
We commend Tooth et al. (1) for their contribution to im-
research in epidemiology. We also appreciate their positive
comments about the Transparent Reporting of Evaluations
with Nonrandomized Designs (TREND) statement (2).
In terms of both content and format, there are strong
similarities among the Consolidated Standards of Reporting
Trials (CONSORT) statement (3), the TREND statement,
and the Tooth et al. paper (1). The checklist and flowchart
approach is clearly gaining recognition as a method for
improving the quality of reporting research. While there
are strong similarities, we would like to call attention to an
important difference in emphasis between TREND and
CONSORT and Tooth et al. The TREND statement asks
authors to describe in detail the theory linking the interven-
tion being reported and the expected outcomes. ‘‘Theory’’
refers to behavioral and/or social science constructs that
link the elements of the intervention to behavioral/social
change processes that would then produce the desired out-
comes. Examples would include cognitive behavior theory
(4) and diffusion of innovations theory (5). A good de-
scription of the underlying theory of the intervention being
evaluated both helps specify process variables for possible
measurement and can increase confidence that the inter-
vention was the cause of the expected outcomes (should
those occur). This can be a major advantage for decision
makers considering implementation of the intervention on
a larger scale.
Within a properly implemented, randomized, clinical trial
design (as reported with CONSORT), there will still be
a relatively strong inference for causation even if the theo-
retical mechanism is unspecified or unknown. In a purely
observational longitudinal study, making causal inferences
is likely to be problematic. However, we suggest adding
to the Tooth et al. checklist (1, p. 283) a theoretical descrip-
tion of the ‘‘exposure’’ that would specify possible causal
linkages between the exposure and the observed outcomes.
Generating cumulative scientific knowledge requires not
only transparent reporting of research studies but also
having theoretical frameworks for integrating the (often
voluminous) data. As we noted in the original TREND
statement paper (2), however, a postanalysis slap-on of a
theoretical framework simply for the sake of having some
theory is not likely to be useful.
Finally, in the spirit of modern science, we also would
like to suggest that Tooth et al. (1) generate a mnemonic
abbreviation for their checklist and flowchart.
Conflict of interest: none declared.
1. Tooth L, Ware R, Bain C, et al. Quality of reporting of obser-
2. Des Jarlais DC, Lyles C, Crepaz N: the TREND Group. Im-
proving the reporting quality of nonrandomized evaluations of
behavioral and public health interventions: the TREND state-
ment. Am J Public Health 2004;94:361–6.
3. Online CONSORT statement. The CONSORT statement.
Strength in science, sound ethics, 2005. (http://www.consort-
4. Bandura A. Social cognitive theory and exercise of control
over HIV infection. In: Peterson J, DiClemente R, eds. Pre-
venting AIDS: theory and practice of behavioral interven-
tions. New York, NY: Plenum Press, 1993.
5. Rogers E. Diffusion of Innovations. New York, NY: The
Free Press, 2003.
Don C. Des Jarlais1, Cynthia M. Lyles2, and Nicole Crepaz1
1Chemical Dependency Institute, Beth Israel Medical
Center, New York, NY 10003
2Centers for Disease Control and Prevention, Atlanta,
DOI: 10.1093/aje/kwi313; Advance Access publication October 5, 2005
1032Am J Epidemiol 2005;162:1032–1035
American Journal of Epidemiology
Copyright ª 2005 by the Johns Hopkins Bloomberg School of Public Health
All rights reserved; printed in U.S.A.
Vol. 162, No. 10
by guest on January 13, 2016